diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 68e69b32b32a8f17fb1c6cea0d85e4867180468c..35b3f778ff7a9ea6571c6197b0196b347eb9d55b 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -2,18 +2,14 @@
image: harbor.containers.wurnet.nl/proxy-cache/library/maven:3.8.4-jdk-8-slim
-stages:
- - test:unit
- - test:documentation
- - test:end-to-end
+variables:
+ MAMBA_ROOT_PREFIX: "/opt/micromamba"
-unit-tests:
- stage: test:unit
- script:
- - mvn test
+stages:
+ - test
sphinx-linter:
- stage: test:documentation
+ stage: test
script:
- |
# Install system dependencies
@@ -30,7 +26,7 @@ sphinx-linter:
sphinx-lint -e=all --max-line-length=80
sphinx-documentation:
- stage: test:documentation
+ stage: test
script:
- |
# Install system dependencies
@@ -46,8 +42,13 @@ sphinx-documentation:
sphinx-build -W docs/source output
-end-to-end-tests:yeast-1:
- stage: test:end-to-end
+unit-tests:
+ stage: test
+ script:
+ - mvn test
+
+end-to-end-yeast-pangenome:
+ stage: test
variables:
KUBERNETES_MEMORY_REQUEST: 8Gi
KUBERNETES_MEMORY_LIMIT: 16Gi
@@ -56,32 +57,35 @@ end-to-end-tests:yeast-1:
# Install system dependencies
apt-get update
- apt-get install --yes --no-install-recommends git
-
- # Install Conda
+ apt-get install --yes --no-install-recommends git bzip2 openssh-client sshpass
- MINICONDA_INSTALLER=$(mktemp)
- curl -o "${MINICONDA_INSTALLER}" --silent "https://repo.anaconda.com/miniconda/Miniconda3-py39_4.12.0-Linux-x86_64.sh"
- bash "${MINICONDA_INSTALLER}" -b
- rm "${MINICONDA_INSTALLER}"
- export PATH="${PATH}:/root/miniconda3/bin/"
+ # Install micromamba
- # Install Mamba for faster installation of remaining packages
+ curl -Ls https://micromamba.snakepit.net/api/micromamba/linux-64/1.4.2 | tar xvj bin/micromamba
+ chmod +x ./bin/micromamba
+ ./bin/micromamba shell --yes init ${MAMBA_ROOT_PREFIX}
+ source /root/.bashrc
- conda install --yes --channel conda-forge "mamba=0.15.3"
+ # Install KMC, samtools and Snakemake with micromamba
- # Install KMC, samtools and Snakemake with Mamba
-
- mamba install --yes --channel conda-forge --channel bioconda \
+ micromamba create --name snakemake --yes --strict-channel-priority \
+ --channel conda-forge --channel bioconda \
"kmc=3.0.1" \
"samtools=1.15" \
"snakemake=7.19.1"
- # Run end-to-end tests with Snakemake
+ # Obtain data from SFTP server
cd tests/
- snakemake \
+ sshpass -e sftp -o StrictHostKeyChecking=no ${SFTP_USERNAME}@${SFTP_HOST} << EOF
+ get /yeast_tiny_pangenome.tar.gz
+ EOF
+ tar xzvf yeast_tiny_pangenome.tar.gz
+
+ # Run end-to-end tests with Snakemake
+
+ micromamba run --name snakemake snakemake \
--cores 1 \
--printshellcmds \
--debug \
- --configfile yeast-1.yaml
+ --configfile yeast.yaml
diff --git a/CHANGELOG.md b/CHANGELOG.md
index f52075d2a66dc19fa6cc48ed4949073b77ca0c86..f89d8499603569999e925011eb6f6435a198917a 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,14 @@
All notable changes to Pantools will be documented in this file.
+## [4.2.2] - 23-06-2023
+
+### Changed
+- `pantools group_info` defaults to all homology groups (!166).
+- `pantools map` no longer accepts a genome numbers file and uses `--include`/`--exclude` instead (!170).
+
+### Fixed
+- `pantools consensus_tree` now properly checks if groups were excluded based on trimming (!168).
+
## [4.2.1] - 05-06-2023
### Added
diff --git a/docs/source/conf.py b/docs/source/conf.py
index 63818a97ddf47d69fb3791c56e1c9b9006562e1c..5349b27201a4e6a2f8271a9212d1901a4c226a07 100644
--- a/docs/source/conf.py
+++ b/docs/source/conf.py
@@ -10,8 +10,8 @@ project = 'PanTools'
copyright = '2016, the PanTools team'
author = 'Sandra Smit'
-release = '4.2.1'
-version = '4.2.1'
+release = '4.2.2'
+version = '4.2.2'
# -- General configuration
diff --git a/docs/source/user_guide/explore.rst b/docs/source/user_guide/explore.rst
index b7873fbf977f2ab4145bf9c7417b5adc32cefb8b..3df1bf337e6d6e14f8827af6dbc606b617f48d08 100644
--- a/docs/source/user_guide/explore.rst
+++ b/docs/source/user_guide/explore.rst
@@ -364,9 +364,6 @@ Report all available information of one or multiple homology groups.
* - <databaseDirectory>
- Path to the database root directory.
- * - <homologyFile>
- - A text file with homology group node identifiers, seperated by a
- comma.
**Options**
.. list-table::
@@ -376,6 +373,14 @@ Report all available information of one or multiple homology groups.
- Only include a selection of genomes.
* - ``--exclude``/``-e``
- Exclude a selection of genomes.
+ * - ``--homology-file``/``-H``
+ - A text file with homology group node identifiers, separated by a comma.
+ Default is all homology groups. (Mutually exclusive with
+ ``--homology-groups``.)
+ * - ``--homology-groups``/``-G``
+ - A comma separated list of homology group node identifiers.
+ Default is all homology groups. (Mutually exclusive with
+ ``--homology-file``.)
* - ``--functions``
- Name of function identifiers from GO, PFAM, InterPro or TIGRAM.
To find Phobius (P) or SignalP (S) annotations, include:
diff --git a/docs/source/user_guide/mapping.rst b/docs/source/user_guide/mapping.rst
index c8abb0a7165ada019ac03c472f16a20819897c14..3da859fef8b2ef9aed66ba087e3e753b67a02879 100644
--- a/docs/source/user_guide/mapping.rst
+++ b/docs/source/user_guide/mapping.rst
@@ -14,9 +14,6 @@ the analysis.
* - <databaseDirectory>
- Path to the database root directory.
- * - <genomeNumbers>
- - A text file containing genome numbers to map reads against in
- each line.
* - <shortReadFiles>
- One or two short-read archives in FASTQ format, which can be
gz/bz2 compressed.
@@ -30,6 +27,10 @@ the analysis.
of cores or 8, whichever is lower.
* - ``--output``/``-o``
- Path to the output files (default is the database path).
+ * - ``--include``/``-i``
+ - Only include a selection of genomes.
+ * - ``--exclude``/``-e``
+ - Exclude a selection of genomes.
* - ``--best-hits`` = ``none|all|random``
- In case of multiple "best" hits, return none, all best hits or a
random best hit (Default: random).
@@ -99,22 +100,14 @@ the analysis.
+
EEEEEEEEEEEEEEEAEEEE/EEEEE/AEEEEEEEEEEEEEE/EE/EEE/<EEEEEEE/EEEEEEEEEEEEEAEEEEEAEEEEEAEEAEEEEEEA<AAAEEAEEA<EE/EEEEAEAEA/EEAA/
- Genome numbers file
-
- .. code:: text
-
- 1
- 2
- 5
-
**Example commands**
.. code:: bash
- $ pantools map arabidopsis_DB genome_numbers.txt ERR031564_1.fastq
- $ pantools map --include=1-5 --sensitivity=sensitive arabidopsis_DB genome_numbers.txt ERR031564_1.fastq
- $ pantools map --competitive -m=all-bests arabidopsis_DB genome_numbers.txt ERR031564_1.fastq
- $ pantools map --interleaved arabidopsis_DB genome_numbers.txt interleaved_reads.fastq
- $ pantools map arabidopsis_DB genome_numbers.txt ERR031564_1.fastq ERR031564_2.fastq
+ $ pantools map arabidopsis_DB ERR031564_1.fastq
+ $ pantools map --include=1-5 --sensitivity=sensitive arabidopsis_DB ERR031564_1.fastq
+ $ pantools map --competitive -m=all-bests arabidopsis_DB ERR031564_1.fastq
+ $ pantools map --interleaved arabidopsis_DB interleaved_reads.fastq
+ $ pantools map arabidopsis_DB ERR031564_1.fastq ERR031564_2.fastq
**Output**
- **mapping_summary.txt**, number of mapped and unmapped reads per
diff --git a/pom.xml b/pom.xml
index 6097914c72911dfaaa78834b9ee711a0d8ac8b16..4b37ce257d8f000b1a432db8936be2e35d8d7f54 100644
--- a/pom.xml
+++ b/pom.xml
@@ -6,7 +6,7 @@
<groupId>nl.wur.bif</groupId>
<artifactId>pantools</artifactId>
- <version>4.2.1</version>
+ <version>4.2.2</version>
<properties>
<maven.compiler.source>8</maven.compiler.source>
diff --git a/src/main/java/nl/wur/bif/pantools/cli/ANI.java b/src/main/java/nl/wur/bif/pantools/cli/ANI.java
index 4f33521957c8135dd8c0e35864273e450393e32c..3c448968492c50f639dfe8680ef5d23eca7721ff 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/ANI.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/ANI.java
@@ -19,7 +19,7 @@ import static picocli.CommandLine.*;
/**
* Calculate Average Nucleotide Identity (ANI) scores between genomes.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "ani", sortOptions = false)
public class ANI implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/AddAnnotations.java b/src/main/java/nl/wur/bif/pantools/cli/AddAnnotations.java
index d0496350a7330530db8deef7a7a9afd3d0163452..7fa5366aebe1b78d2e67cb0ec218bdbe1f7ef0a6 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/AddAnnotations.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/AddAnnotations.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.*;
/**
* Adds genomic features to genomes and generates their proteomes.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "add_annotations", sortOptions = false)
public class AddAnnotations implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/AddAntiSMASH.java b/src/main/java/nl/wur/bif/pantools/cli/AddAntiSMASH.java
index 60cc9bef58ee0547e3d2428241d631718d4689ae..0c0fca2a77e96716c878536871d429cbedbad1c0 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/AddAntiSMASH.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/AddAntiSMASH.java
@@ -16,7 +16,7 @@ import static nl.wur.bif.pantools.utils.Globals.*;
/**
* Include antiSMASH gene clusters into the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "add_antismash", sortOptions = false)
public class AddAntiSMASH implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/AddFunctions.java b/src/main/java/nl/wur/bif/pantools/cli/AddFunctions.java
index a9cc98b965a6f2a44e87c788a25592318eace0ce..63cc1f5cf46b5ea3906053909a98dff6b7e5a03f 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/AddFunctions.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/AddFunctions.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.*;
/**
* Add functional annotations to a pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "add_functions", sortOptions = false)
public class AddFunctions implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/AddGenomes.java b/src/main/java/nl/wur/bif/pantools/cli/AddGenomes.java
index 55b3a572e4e278ccde7a57b16176cafb93d7348f..1c2536e44e54f69e6752bd935840f40a3a147269 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/AddGenomes.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/AddGenomes.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Include additional genomes to an already available pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "add_genomes", sortOptions = false)
public class AddGenomes implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/AddPhenotypes.java b/src/main/java/nl/wur/bif/pantools/cli/AddPhenotypes.java
index e862bbbbb294c04409116758720b51fda69315ed..ad3044b9d5472a1d3ad4926582b278f4b95d0578 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/AddPhenotypes.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/AddPhenotypes.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Include phenotype data to the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "add_phenotypes", sortOptions = false)
public class AddPhenotypes implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/BuildPangenome.java b/src/main/java/nl/wur/bif/pantools/cli/BuildPangenome.java
index 58aee792ddbf37b6e9ed6d0294a1c695bd7b3e9c..a1b97d30ed294502a29f0018589c99194f7867c4 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/BuildPangenome.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/BuildPangenome.java
@@ -19,7 +19,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Build a pangenome out of a set of genomes.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "build_pangenome", sortOptions = false)
public class BuildPangenome implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/BuildPanproteome.java b/src/main/java/nl/wur/bif/pantools/cli/BuildPanproteome.java
index 9e837df17446bc2d25b0e6c66907511434f631b4..91e27e5a108f0bc3d41f49509b35231a6bcb71b3 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/BuildPanproteome.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/BuildPanproteome.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Build a panproteome out of a set of proteins.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "build_panproteome", sortOptions = false)
public class BuildPanproteome implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/BuscoProtein.java b/src/main/java/nl/wur/bif/pantools/cli/BuscoProtein.java
index 5ae9ceb556dce1bf6f7c04465cfe90945c838e7c..2b4645288e47dd182e931336969f4e4a77e6ec56 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/BuscoProtein.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/BuscoProtein.java
@@ -22,7 +22,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Identify BUSCO genes in the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "busco_protein", sortOptions = false, abbreviateSynopsis = true)
public class BuscoProtein implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/ChangeGrouping.java b/src/main/java/nl/wur/bif/pantools/cli/ChangeGrouping.java
index 42a7aa87d08a9f866f5474f7666fe352558a68b1..66f2f70f2da41873c6cac3b626b5c73dae73fd1b 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/ChangeGrouping.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/ChangeGrouping.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Change the active version of the homology grouping.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "change_grouping", sortOptions = false)
public class ChangeGrouping implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/CompareGO.java b/src/main/java/nl/wur/bif/pantools/cli/CompareGO.java
index 32aa1aba7672ce8a8f50463b1f1768087f5c538d..4f8dd270b407d8ec87758d063268703b58b5fa37 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/CompareGO.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/CompareGO.java
@@ -21,7 +21,7 @@ import static picocli.CommandLine.*;
/**
* For two given GO terms, move up in the GO hierarchy to see if they are related.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "compare_go", sortOptions = false)
public class CompareGO implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/ConsensusTree.java b/src/main/java/nl/wur/bif/pantools/cli/ConsensusTree.java
index f67b732a88b2c3cb47dc75cabec3a0fcfeafba06..c2319a9e13ba0c70555aee4fcfc0919d1ea0903b 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/ConsensusTree.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/ConsensusTree.java
@@ -18,7 +18,7 @@ import static picocli.CommandLine.*;
/**
* Create a consensus tree by combining gene trees from homology groups using ASTRAL-Pro.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "consensus_tree", sortOptions = false)
public class ConsensusTree implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/CorePhylogeny.java b/src/main/java/nl/wur/bif/pantools/cli/CorePhylogeny.java
index 620d6fb93580bb8a110b118c09c6e2c76e2a4e3a..0af39d68a869b876ffa5c3b544e47f8dd8dab33a 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/CorePhylogeny.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/CorePhylogeny.java
@@ -21,7 +21,7 @@ import static picocli.CommandLine.*;
/**
* Create a SNP tree from single-copy genes.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "core_phylogeny", aliases = "core_snp_tree", sortOptions = false)
public class CorePhylogeny implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/CoreUniqueThresholds.java b/src/main/java/nl/wur/bif/pantools/cli/CoreUniqueThresholds.java
index bba3aa8fffc2f60ebda0d94ae11835ee05c434ab..0eaf10e84c87e429ae01002742f1287857171b73 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/CoreUniqueThresholds.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/CoreUniqueThresholds.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.*;
/**
* Test the effect of changing the core and unique threshold.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "core_unique_thresholds", sortOptions = false)
public class CoreUniqueThresholds implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/CreateTreeTemplate.java b/src/main/java/nl/wur/bif/pantools/cli/CreateTreeTemplate.java
index 50f101f9c32bbff522c6bcfd872c38723a4098fe..381f2055b0e91cea9a9abd13e09d9495ecd2f394 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/CreateTreeTemplate.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/CreateTreeTemplate.java
@@ -15,7 +15,7 @@ import static picocli.CommandLine.*;
/**
* Create templates for coloring phylogenetic trees in iTOL.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "create_tree_template", sortOptions = false)
public class CreateTreeTemplate implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/ExportPangenome.java b/src/main/java/nl/wur/bif/pantools/cli/ExportPangenome.java
index 92cae74ad52d18fd81aaa2afe20377658f910e4e..33b8341cdc71d7629e032a2fd38a9d67401326f4 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/ExportPangenome.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/ExportPangenome.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.*;
/**
* Classify functional annotations as core, accessory or unique.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@CommandLine.Command(name = "export_pangenome", sortOptions = false, abbreviateSynopsis = true)
public class ExportPangenome implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/FindGenesByAnnotation.java b/src/main/java/nl/wur/bif/pantools/cli/FindGenesByAnnotation.java
index 20b30d7720f6c74b58091837c7367c7208ef902c..2cf4d42382cc36bbc0bd71c3d63aeb89e3a3b201 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/FindGenesByAnnotation.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/FindGenesByAnnotation.java
@@ -21,7 +21,7 @@ import static picocli.CommandLine.*;
* Find genes of interest in the pangenome that share a functional annotation node and extract the nucleotide and
* protein sequence.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "find_genes_by_annotation", sortOptions = false)
public class FindGenesByAnnotation implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/FindGenesByName.java b/src/main/java/nl/wur/bif/pantools/cli/FindGenesByName.java
index d1d2235576cbee6ecef285136bea783230b1b5cf..1a2deba4580003869b7b731a7491094c9e067016 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/FindGenesByName.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/FindGenesByName.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.*;
/**
* Find your genes of interest in the pangenome by using the gene name and extract the nucleotide and protein sequence.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "find_genes_by_name", sortOptions = false)
public class FindGenesByName implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/FindGenesInRegion.java b/src/main/java/nl/wur/bif/pantools/cli/FindGenesInRegion.java
index 12e143d22b0e3f65ab3964bc707df6d39d424ca6..0c715ce6fb79ac018bb19be8b3aad37244f5d162 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/FindGenesInRegion.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/FindGenesInRegion.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.*;
/**
* Find genes in a given genomic region.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "find_genes_in_region", sortOptions = false)
public class FindGenesInRegion implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/FunctionOverview.java b/src/main/java/nl/wur/bif/pantools/cli/FunctionOverview.java
index 99e5ef9d3072ee2ef89780acc084aa6b74e33ee5..3b8f60853ae504ed3d9c496a7ca3c12d78573de4 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/FunctionOverview.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/FunctionOverview.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.*;
/**
* Create an overview table for each functional annotation type in the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "function_overview", sortOptions = false)
public class FunctionOverview implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/FunctionalClassification.java b/src/main/java/nl/wur/bif/pantools/cli/FunctionalClassification.java
index 9b09de7de7e1b805e7e2650191eb1c8db1786e44..130d284630a753bb12d9d38cb66c0acf648b5e28 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/FunctionalClassification.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/FunctionalClassification.java
@@ -19,7 +19,7 @@ import static picocli.CommandLine.*;
/**
* Classify functional annotations as core, accessory or unique.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "functional_classification", sortOptions = false, abbreviateSynopsis = true)
public class FunctionalClassification implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/GOEnrichment.java b/src/main/java/nl/wur/bif/pantools/cli/GOEnrichment.java
index 29e31b6ff892a1751c5ebeb71584822a66631110..f0e2b9c0262d268aa4be1a74407f2fc37ee259de 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/GOEnrichment.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/GOEnrichment.java
@@ -22,7 +22,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Identify over or underrepresented GO terms in a set of genes.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "go_enrichment", sortOptions = false)
public class GOEnrichment implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/GeneClassification.java b/src/main/java/nl/wur/bif/pantools/cli/GeneClassification.java
index 595d0c0473fa3cef93b97e95b9f85d833956f5f4..c10fc40f7fa3f33e87d9773fdcd8242f00bae8c5 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/GeneClassification.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/GeneClassification.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Classify the gene repertoire as core, accessory or unique.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "gene_classification", sortOptions = false, abbreviateSynopsis = true)
public class GeneClassification implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/Group.java b/src/main/java/nl/wur/bif/pantools/cli/Group.java
index 33a6dd6b4a9e70150d1896360cb78fad0fa66fdc..0339c167b5a5e0f7c46af0cc20666dee47699bc8 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/Group.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/Group.java
@@ -23,7 +23,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Generate homology groups based on similarity of protein sequences.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "group", sortOptions = false, abbreviateSynopsis = true)
public class Group implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/GroupInfo.java b/src/main/java/nl/wur/bif/pantools/cli/GroupInfo.java
index 690bd30cc957a4969277124ed797d872689e47eb..281493724fe070026ae64c138237886a467e168f 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/GroupInfo.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/GroupInfo.java
@@ -18,14 +18,14 @@ import static picocli.CommandLine.*;
/**
* Report all available information of one or multiple homology groups.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "group_info", sortOptions = false)
public class GroupInfo implements Callable<Integer> {
@Spec Model.CommandSpec spec;
@ArgGroup private SelectGenomes selectGenomes;
- @ArgGroup(multiplicity = "1") private SelectHmGroups selectHmGroups;
+ @ArgGroup private SelectHmGroups selectHmGroups = new SelectHmGroups();
@ParentCommand
@GraphDatabase
@@ -61,7 +61,6 @@ public class GroupInfo implements Callable<Integer> {
private void setGlobalParameters() throws IOException {
setGenomeSelectionOptions(selectGenomes);
-// SELECTED_HMGROUPS = selectHmGroups.getHomologyGroups().stream().map(String::valueOf).collect(Collectors.joining(","));
if (genes != null) SELECTED_NAME = genes.toString().replaceAll("[\\[\\]]", "");
if (functions != null) SELECTED_LABEL = functions.toString().replaceAll("[\\[\\]]", "");
diff --git a/src/main/java/nl/wur/bif/pantools/cli/GroupingOverview.java b/src/main/java/nl/wur/bif/pantools/cli/GroupingOverview.java
index b98e32882ac32b6c35c1e2a73ffc5a6d66af7fae..b42b4183e3ca3ce846ed6aa67fd9ab7718a0854c 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/GroupingOverview.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/GroupingOverview.java
@@ -15,7 +15,7 @@ import static picocli.CommandLine.*;
/**
* Create an overview table for every homology grouping in the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "grouping_overview", sortOptions = false)
public class GroupingOverview implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/KmerClassification.java b/src/main/java/nl/wur/bif/pantools/cli/KmerClassification.java
index fc1b2156d8e5a202c05874c5f6d9c02cad4d21b4..b6e967fed511f77dff8b8bf737c4e97039eafa66 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/KmerClassification.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/KmerClassification.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Calculate the number of core, accessory, unique, k-mer sequences.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "k_mer_classification", aliases = "kmer_classification", sortOptions = false, abbreviateSynopsis = true)
public class KmerClassification implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/LocateGenes.java b/src/main/java/nl/wur/bif/pantools/cli/LocateGenes.java
index 1577737f69458112f01b10dd4f17565035da722b..398cfe7f3a7e4579f1c15ef8a94daa01fa3eb47f 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/LocateGenes.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/LocateGenes.java
@@ -21,7 +21,7 @@ import static picocli.CommandLine.*;
/**
* Identify and compare gene clusters of from a set of homology groups.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "locate_genes", sortOptions = false, abbreviateSynopsis = true)
public class LocateGenes implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/MLSA.java b/src/main/java/nl/wur/bif/pantools/cli/MLSA.java
index 517de6278fed9bab444b8f5835410b7132943fc7..0b6410e040900cf55301215a25dd8efacce0727c 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/MLSA.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/MLSA.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.*;
/**
* Step 3/3 of MLSA. Run IQ-tree on the concatenated sequences.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "mlsa", sortOptions = false)
public class MLSA implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/MLSAConcatenate.java b/src/main/java/nl/wur/bif/pantools/cli/MLSAConcatenate.java
index 14beadae095b88939f6104f8a62e4f48a655074a..0a96cc63e384a50343ad878483f20e425e31d1ca 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/MLSAConcatenate.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/MLSAConcatenate.java
@@ -18,7 +18,7 @@ import static picocli.CommandLine.*;
/**
* Step 2/3 of MLSA. Concatenate the gene selection into a single continuous sequence.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "mlsa_concatenate", sortOptions = false)
public class MLSAConcatenate implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/MLSAFindGenes.java b/src/main/java/nl/wur/bif/pantools/cli/MLSAFindGenes.java
index 438c30a323b1b45393522feead99739177160dcd..4438c2eb23a1ff4dfd574175a724ce05d878bba2 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/MLSAFindGenes.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/MLSAFindGenes.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Step 1/3 of MLSA. Search and filter suitable genes for the MLSA.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "mlsa_find_genes", sortOptions = false)
public class MLSAFindGenes implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/MSA.java b/src/main/java/nl/wur/bif/pantools/cli/MSA.java
index 6936fdacbdd6c49000810621e479b865688e1f73..18911e030ea26b075faae8e4346960fa57375772 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/MSA.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/MSA.java
@@ -26,7 +26,7 @@ import static picocli.CommandLine.*;
/**
* Create multiple sequence alignments.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "msa", sortOptions = false, abbreviateSynopsis = true)
public class MSA implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/Map.java b/src/main/java/nl/wur/bif/pantools/cli/Map.java
index de801d48c490856032280c08e84b877eab09f710..8d58dec14a1fd7cf0fbb1f79ed1b994bb361f4be 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/Map.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/Map.java
@@ -1,6 +1,7 @@
package nl.wur.bif.pantools.cli;
import jakarta.validation.constraints.*;
+import nl.wur.bif.pantools.cli.mixins.SelectGenomes;
import nl.wur.bif.pantools.cli.mixins.ThreadNumber;
import nl.wur.bif.pantools.utils.BeanUtils;
import nl.wur.bif.pantools.pantools.Pantools;
@@ -19,7 +20,7 @@ import static picocli.CommandLine.*;
/**
* Map single or paired-end short reads to one or multiple genomes in the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "map", sortOptions = false, abbreviateSynopsis = true)
@OutputDirectory(directory = "outDirectory")
@@ -27,15 +28,12 @@ public class Map implements Callable<Integer> {
@Spec static CommandSpec spec;
@Mixin private ThreadNumber threadNumber;
+ @ArgGroup private SelectGenomes selectGenomes;
@ParentCommand
@GraphDatabase
private Pantools pantools;
- @Parameters(descriptionKey = "genome-numbers", index = "0+")
- @InputFile(message = "{file.genome-numbers}")
- Path genomeNumbers;
-
@Parameters(descriptionKey = "short-read-file", index = "1+", arity = "1..2")
@InputFiles(message = "{file.short-read}")
Path[] shortReadFiles;
@@ -204,7 +202,7 @@ public class Map implements Callable<Integer> {
}
private void setGlobalParameters() {
- PATH_TO_THE_GENOME_NUMBERS_FILE = genomeNumbers.toString();
+ setGenomeSelectionOptions(selectGenomes);
PATH_TO_THE_FIRST_SRA = shortReadFiles[0].toString();
if (shortReadFiles.length == 2) PATH_TO_THE_SECOND_SRA = shortReadFiles[1].toString();
if (previousRunFile != null) RAW_ABUNDANCE_FILE = previousRunFile.toString();
diff --git a/src/main/java/nl/wur/bif/pantools/cli/Metrics.java b/src/main/java/nl/wur/bif/pantools/cli/Metrics.java
index 8920c56d4adc7e19b32d46952febf3b939773e8c..bc1ac60b3097585214d953efb5bb592d58d0d780 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/Metrics.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/Metrics.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.*;
/**
* Generates relevant metrics of the pangenome and the individual genomes and sequences.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "metrics", sortOptions = false)
public class Metrics implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/MoveGrouping.java b/src/main/java/nl/wur/bif/pantools/cli/MoveGrouping.java
index 1d5c1607cd709760d97c6d386a22c8f77a5f838e..79972424823be3fe9927eaa8a574b235d9d4bfcd 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/MoveGrouping.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/MoveGrouping.java
@@ -15,7 +15,7 @@ import static picocli.CommandLine.*;
/**
* Inactivate the currently active homology grouping.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "move_grouping", sortOptions = false)
public class MoveGrouping implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/OptimalGrouping.java b/src/main/java/nl/wur/bif/pantools/cli/OptimalGrouping.java
index aa8991ffef07dace734575dfe37a9021c285c83b..c22de667a8816c8414422791f17686cbffd4581b 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/OptimalGrouping.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/OptimalGrouping.java
@@ -22,7 +22,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Find the most suitable settings for <Group>.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "optimal_grouping", sortOptions = false)
public class OptimalGrouping implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/OrderMatrix.java b/src/main/java/nl/wur/bif/pantools/cli/OrderMatrix.java
index 14a7e6d970e31555dce447a6cdfe73872f281474..2e457a84f2264ef4b150cf976d4b0e352cf3a748 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/OrderMatrix.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/OrderMatrix.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.*;
/**
* Order the values of a matrix file created by PanTools.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "order_matrix", sortOptions = false)
public class OrderMatrix implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/PangenomeStructure.java b/src/main/java/nl/wur/bif/pantools/cli/PangenomeStructure.java
index d381296070f98214bf3033a9a2407f53a84e211c..11ebfe0ef6bfe87749b1ce235fcf275485307917 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/PangenomeStructure.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/PangenomeStructure.java
@@ -18,7 +18,7 @@ import static picocli.CommandLine.*;
/**
* Determine the openness of the pangenome based on k-mer sequences.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "pangenome_structure", sortOptions = false, abbreviateSynopsis = true)
public class PangenomeStructure implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RemoveAnnotations.java b/src/main/java/nl/wur/bif/pantools/cli/RemoveAnnotations.java
index 7048f0b4d6bb42d365654df80f35683632e24817..3ce09ffd9c04550ade6d3513945714b33f377ba9 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RemoveAnnotations.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RemoveAnnotations.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Remove all the genomic features that belong to annotations.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "remove_annotations", sortOptions = false)
public class RemoveAnnotations implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RemoveFunctions.java b/src/main/java/nl/wur/bif/pantools/cli/RemoveFunctions.java
index c2e2d33845a9a1f5b5bdeb83ba474f0b7bcdd523..85ff4a7efa819c03fb01b1e9e07666c7f65ab173 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RemoveFunctions.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RemoveFunctions.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.*;
/**
* Remove an homology grouping from the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "remove_functions", sortOptions = false)
public class RemoveFunctions implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RemoveGrouping.java b/src/main/java/nl/wur/bif/pantools/cli/RemoveGrouping.java
index 463b9bd74d82b877336d76acef8fad9110250f02..ab70a0e87b6482d844a1ee0916522e38b6def5e0 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RemoveGrouping.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RemoveGrouping.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.*;
/**
* Remove an homology grouping from the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "remove_grouping", sortOptions = false)
public class RemoveGrouping implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RemoveNodes.java b/src/main/java/nl/wur/bif/pantools/cli/RemoveNodes.java
index 63239c9ea608e1dd9a1cf843bfc7b71814da33c8..cb94da121231f7719967fe4dc76ce38479799202 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RemoveNodes.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RemoveNodes.java
@@ -19,7 +19,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Remove a selection of nodes and their relationships from the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "remove_nodes", sortOptions = false)
@ExcludePatterns(regexp = "nucleotide,sequence,pangenome,genome,degenerate", field = "label",
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RemovePhenotypes.java b/src/main/java/nl/wur/bif/pantools/cli/RemovePhenotypes.java
index 056370da0d95175882294b495cb391bcb998b737..3c6202eef8553a3903c6071be00ebe798fd36e78 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RemovePhenotypes.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RemovePhenotypes.java
@@ -15,7 +15,7 @@ import static picocli.CommandLine.*;
/**
* Delete phenotype nodes or remove specific phenotype information from the nodes.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "remove_phenotypes", sortOptions = false)
public class RemovePhenotypes implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RenameMatrix.java b/src/main/java/nl/wur/bif/pantools/cli/RenameMatrix.java
index f022b53501a33e44f01b241d5ece7980f1c5742f..c3d182daea437736825c4d3ae25942e6970af536 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RenameMatrix.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RenameMatrix.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.*;
/**
* Rename the headers of CSV formatted matrix files.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "rename_matrix", sortOptions = false)
public class RenameMatrix implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RenamePhylogeny.java b/src/main/java/nl/wur/bif/pantools/cli/RenamePhylogeny.java
index bbd9ea27cc7b44f2a9477416b85ea41c3eba13c2..cd81156b7472c21dbce03c8379b3c91f8646e4a4 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RenamePhylogeny.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RenamePhylogeny.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* Update or alter the terminal nodes (leaves) of a phylogenic tree.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "rename_phylogeny")
public class RenamePhylogeny implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RetrieveFeatures.java b/src/main/java/nl/wur/bif/pantools/cli/RetrieveFeatures.java
index ed09de21fa7292126eacb8bfa1ca62d1ccfd4f2f..79139cfee8dded276332ba5b8edcf701ece2547c 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RetrieveFeatures.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RetrieveFeatures.java
@@ -15,7 +15,7 @@ import static picocli.CommandLine.*;
/**
* Retrieve the sequence of annotated features from the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "retrieve_features", sortOptions = false)
public class RetrieveFeatures implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RetrieveRegions.java b/src/main/java/nl/wur/bif/pantools/cli/RetrieveRegions.java
index db50570ab9f217b7b9050cc30ffa8ed4883e709c..31867590e399948c77fdf8f216b7626034fabc28 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RetrieveRegions.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RetrieveRegions.java
@@ -17,7 +17,7 @@ import static picocli.CommandLine.*;
/**
* Retrieve the sequence of genomic regions from the pangenome.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "retrieve_regions", sortOptions = false)
public class RetrieveRegions implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/RootPhylogeny.java b/src/main/java/nl/wur/bif/pantools/cli/RootPhylogeny.java
index d2b7a05676fcbaf972e92a3ff0d295ad8a6292a3..1cd092fb7acdc6ba37c45278f5ec78b8909f47b5 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/RootPhylogeny.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/RootPhylogeny.java
@@ -16,7 +16,7 @@ import static picocli.CommandLine.Model.CommandSpec;
/**
* (Re)root a phylogenetic tree.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "root_phylogeny", sortOptions = false)
public class RootPhylogeny implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/ShowGO.java b/src/main/java/nl/wur/bif/pantools/cli/ShowGO.java
index 97bf8b9303aadaf2c15d799dc84b91ad73c5184b..e0babec840fd7dcb32a1e9ecfca0c7a87f0fe68d 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/ShowGO.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/ShowGO.java
@@ -19,7 +19,7 @@ import static picocli.CommandLine.*;
/**
* For a given GO term, show the child terms, all parent terms higher in the hierarchy, and connected mRNA nodes.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
@Command(name = "show_go", sortOptions = false)
public class ShowGO implements Callable<Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/mixins/SelectGenomes.java b/src/main/java/nl/wur/bif/pantools/cli/mixins/SelectGenomes.java
index 402adeb0cd7591c429fc08682a0af5c1354fc4c1..8f05de2be09913ff0e847d8699c36ae86c61e695 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/mixins/SelectGenomes.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/mixins/SelectGenomes.java
@@ -12,7 +12,7 @@ import static picocli.CommandLine.*;
* Argument group (mutually exclusive) for '--include' and '--exclude' options. Accepts lists (1,2,3), ranges (5-10)
* or a combination (1,2,3,5-10) as user input for both options.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class SelectGenomes {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/mixins/ThreadNumber.java b/src/main/java/nl/wur/bif/pantools/cli/mixins/ThreadNumber.java
index e2cc63504a77358df4db40a44055fdb919b95726..ab92b161d59016d1e3908679c02890a962f05c01 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/mixins/ThreadNumber.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/mixins/ThreadNumber.java
@@ -7,7 +7,7 @@ import static picocli.CommandLine.Option;
/**
* Mixin class for '--threads' option.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class ThreadNumber {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/Constraints.java b/src/main/java/nl/wur/bif/pantools/cli/validation/Constraints.java
index 0d28c9bf9f58c2db431a9999b0236fb7c678699d..bb1fb05d05c9fb5831765d55acab8de6f0efbd23 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/Constraints.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/Constraints.java
@@ -12,7 +12,7 @@ import static java.lang.annotation.RetentionPolicy.RUNTIME;
/**
* Class containing custom Bean validation constraint interfaces.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class Constraints {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/ExcludePatternsValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/ExcludePatternsValidator.java
index 67f5d60723abe58ade236fa773d0b0b3212e5d62..b4fb78b6e41dca3a9d07bbe1e9f854db8cde3a45 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/ExcludePatternsValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/ExcludePatternsValidator.java
@@ -20,7 +20,7 @@ import static java.util.regex.Pattern.compile;
* Custom ConstraintValidator that verifies whether an input string matches 'forbidden' words. Allows queries to enable
* yes|no user input for continuing after encountering a forbidden word.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class ExcludePatternsValidator implements ConstraintValidator<ExcludePatterns, Object> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputDirectoryValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputDirectoryValidator.java
index 9255ed33175ad677ad354fc1733f12358c7bdf14..90544d58425f5d5a87a7901fbf6e44964ccae26d 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputDirectoryValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputDirectoryValidator.java
@@ -10,7 +10,7 @@ import java.nio.file.Path;
/**
* Custom ConstraintValidator that verifies that an input directory exists and is a directory.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class InputDirectoryValidator implements ConstraintValidator<InputDirectory, Path> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputFileValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputFileValidator.java
index e4b2aa7cf441ff14a35f62e1040e2c8a028ed4cd..02394c2081dd5431ba6c11c9110d4f191dac4f70 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputFileValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputFileValidator.java
@@ -12,7 +12,7 @@ import static nl.wur.bif.pantools.utils.BeanUtils.setViolationMessage;
/**
* Custom ConstraintValidator that verifies that an input file exists.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class InputFileValidator implements ConstraintValidator<InputFile, Path> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputFilesValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputFilesValidator.java
index 469f18845db789beee0fa366fdbe00ccc4d8019d..df3415f92e9cb442282f81e4874318d8e1b02f2e 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputFilesValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/InputFilesValidator.java
@@ -12,7 +12,7 @@ import static nl.wur.bif.pantools.utils.BeanUtils.setViolationMessage;
/**
* Custom ConstraintValidator that verifies that all files in an array of input file exist.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class InputFilesValidator implements ConstraintValidator<InputFiles, Path[]> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/MatchIntegerValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/MatchIntegerValidator.java
index 6aee85b0589ba1647ac7bc91f953b46d54784cf7..c8a11b7867ae47a4b20cf04eb01d536bef2e7db6 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/MatchIntegerValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/MatchIntegerValidator.java
@@ -9,7 +9,7 @@ import jakarta.validation.ConstraintValidatorContext;
/**
* Custom ConstraintValidator that matches an input integer to a selection of allowed options.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class MatchIntegerValidator implements ConstraintValidator<MatchInteger, Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/MinOrZeroValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/MinOrZeroValidator.java
index 1a6ae670fb45b84ae66b9291c170f3d1438a20ba..6dacce9ccc6c38c640075bd702aa9d301a0a20c6 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/MinOrZeroValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/MinOrZeroValidator.java
@@ -8,7 +8,7 @@ import jakarta.validation.ConstraintValidatorContext;
/**
* Custom version of the @Min constraint for optional integers that allows 0 as a valid option.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class MinOrZeroValidator implements ConstraintValidator<MinOrZero, Integer> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/OutputDirectoryValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/OutputDirectoryValidator.java
index 476f380ac4d848cb4f0b07ab5399df3b8f3d634c..0f4ec676e7095841b39353cf247925e71410595c 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/OutputDirectoryValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/OutputDirectoryValidator.java
@@ -18,7 +18,7 @@ import static nl.wur.bif.pantools.cli.validation.Constraints.*;
* Can validate the @ParentCommand decorated field 'pantools' when the given directory is 'databaseDirectory'.
* The subcommand needs getters for the Pantools parent class and the output directory path.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class OutputDirectoryValidator implements ConstraintValidator<OutputDirectory, Object> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PantoolsDatabaseValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PantoolsDatabaseValidator.java
index 1a8ec1c6a9751c9ffa05398121920705d1d577c0..6f2969b1583b991d2e83768eead558ec3be84a69 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PantoolsDatabaseValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PantoolsDatabaseValidator.java
@@ -15,7 +15,7 @@ import static nl.wur.bif.pantools.utils.BeanUtils.setViolationMessage;
/**
* Custom ConstraintValidator that validates that the database directory exists and has an existing neo4j graph database.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class PantoolsDatabaseValidator implements ConstraintValidator<GraphDatabase, Pantools> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PatternListValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PatternListValidator.java
index fe18888e36356e6fe5162613d638283f73669683..d8c1f8e57e70d9782b4af9f0d7519028fcc8c41e 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PatternListValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PatternListValidator.java
@@ -15,7 +15,7 @@ import static nl.wur.bif.pantools.cli.validation.Constraints.*;
/**
* Custom version of @Pattern constraint that matches a list of strings with a regular expression.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class PatternListValidator implements ConstraintValidator<Patterns, List<String>> {
diff --git a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PatternMapValidator.java b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PatternMapValidator.java
index 0604a36c6679f5dbfd4720fd7c7fc23633662d14..b0be86928710e5ab5064cc62435997f65cbaa5e0 100644
--- a/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PatternMapValidator.java
+++ b/src/main/java/nl/wur/bif/pantools/cli/validation/validators/PatternMapValidator.java
@@ -15,7 +15,7 @@ import static nl.wur.bif.pantools.cli.validation.Constraints.Patterns;
/**
* Custom version of @Pattern constraint that matches a list of strings with a regular expression that matches a keyset of strings.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public class PatternMapValidator implements ConstraintValidator<Patterns, HashMap<String, String>> {
diff --git a/src/main/java/nl/wur/bif/pantools/pangenome/AnnotationLayer.java b/src/main/java/nl/wur/bif/pantools/pangenome/AnnotationLayer.java
index 8c24b390ab767eadb51b8d3f7994ef5f764784a6..810c5452678434d535fea5eb0b9977955edb88c5 100755
--- a/src/main/java/nl/wur/bif/pantools/pangenome/AnnotationLayer.java
+++ b/src/main/java/nl/wur/bif/pantools/pangenome/AnnotationLayer.java
@@ -1515,11 +1515,8 @@ public class AnnotationLayer {
Pantools.logger.warn("Allowed features: gene, coding_gene, mRNA, CDS, exon, intron, tRNA, rRNA");
}
- // Verify if the genomes were correctly selected
- if (PATH_TO_THE_GENOME_NUMBERS_FILE != null && (skip_genomes != null || target_genome != null)) {
- Pantools.logger.error("Select genomes via --feature-type, --skip or --reference. Using multiple arguments is not allowed.");
- throw new RuntimeException("Too many genome selection arguments");
- } else if (PATH_TO_THE_GENOME_NUMBERS_FILE == null && skip_genomes == null && target_genome == null) {
+ // Verify if the genomes were correctly selected
+ if (skip_genomes == null && target_genome == null) {
Pantools.logger.error("Missing a genome selection via --feature-type, --skip or --reference. Using multiple arguments is not allowed.");
throw new RuntimeException("Missing genome selection argument");
}
@@ -1543,13 +1540,6 @@ public class AnnotationLayer {
// When using --label, the array is prepared below
BufferedWriter[] out = new BufferedWriter[GENOME_DB.num_genomes];
- if (PATH_TO_THE_GENOME_NUMBERS_FILE != null) { // skip_array is prepared
- skip_array = new boolean[total_genomes];
- TreeSet<Integer> genome_numbers_set = retrieve_numbers_from_genome_numbers_file();
- for (int i = 1; i <= total_genomes; i++) {
- skip_array[i-1] = !genome_numbers_set.contains(i); // sets true or false in array
- }
- }
try { // start the file writers
for (int i = 0; i < total_genomes; i++) { // i+1 is the genome number
@@ -1559,7 +1549,7 @@ public class AnnotationLayer {
out[i] = new BufferedWriter(new FileWriter(WORKING_DIRECTORY + "retrieval/features/" + SELECTED_LABEL + "s_" + (i+1) + ".fasta"));
}
} catch (IOException ioe) {
- Pantools.logger.error("Failed to read: {}", PATH_TO_THE_GENOME_NUMBERS_FILE);
+ Pantools.logger.error("Failed to read fasta file.");
throw new RuntimeException("File error");
}
@@ -1605,36 +1595,6 @@ public class AnnotationLayer {
Pantools.logger.info("Output written to:");
Pantools.logger.info("{}retrieval/features/", WORKING_DIRECTORY);
}
-
- /**
- * Read --genome-numbers or -gn. Places the numbers in a Set that automatically sorts the numbers in ascending order
- * @return
- */
- public TreeSet<Integer> retrieve_numbers_from_genome_numbers_file() {
- String line;
- TreeSet<Integer> genome_numbers = new TreeSet<>();
- try {
- BufferedReader in; in = new BufferedReader(new FileReader(PATH_TO_THE_GENOME_NUMBERS_FILE));
- while ((line = in.readLine()) != null) {
- line = line.trim();
- if (line.equals("")) {
- continue;
- }
- try {
- int genome = Integer.parseInt(line.trim());
- genome_numbers.add(genome);
- } catch(NumberFormatException e) {
- Pantools.logger.info("{}is not a valid genome number.", line.trim());
- }
-
- }
- in.close();
- } catch (IOException e) {
- Pantools.logger.error(e.getMessage());
- System.exit(1);
- }
- return genome_numbers;
- }
/**
* Translates a transcript to a protein sequence
diff --git a/src/main/java/nl/wur/bif/pantools/pangenome/Classification.java b/src/main/java/nl/wur/bif/pantools/pangenome/Classification.java
index 50f4dd70dcc0a1b3211ecad5a9ffd7855aac9ab0..967764310143dfe15aec9837e721d5fbccfa501a 100644
--- a/src/main/java/nl/wur/bif/pantools/pangenome/Classification.java
+++ b/src/main/java/nl/wur/bif/pantools/pangenome/Classification.java
@@ -1036,8 +1036,9 @@ public class Classification {
* Optional
* -label
*/
- public void homology_group_info(List<Long> homologyGroups) throws RuntimeException {
+ public void homology_group_info(List<Long> homologyGroups) throws RuntimeException, IOException {
Pantools.logger.info("Reporting all information from selected homology groups.");
+
StringBuilder output_builder = new StringBuilder();
HashSet<String> groups_with_function = new HashSet<>();
HashMap<String, HashSet<String>> groups_with_function_or_name = new HashMap<>();
@@ -8543,7 +8544,7 @@ public class Classification {
if (gene_or_kmer.equals("kmer")) {
rscript.append(create_pangenome_size_rscript_kmer(""));
- rscript.append(create_pangenome_size_rscript_kmer("unique"));
+ rscript.append(create_pangenome_size_rscript_kmer("_unique"));
} else if (gene_or_kmer.equals("gene")) {
rscript.append(create_pangenome_size_rscript_gene());
} else {
diff --git a/src/main/java/nl/wur/bif/pantools/pangenome/GenomeLayer.java b/src/main/java/nl/wur/bif/pantools/pangenome/GenomeLayer.java
index da46bd2747c81a85edf4d22adb12cac94059c912..b1fb25ec235b1b795876bb562e3258482e30f325 100755
--- a/src/main/java/nl/wur/bif/pantools/pangenome/GenomeLayer.java
+++ b/src/main/java/nl/wur/bif/pantools/pangenome/GenomeLayer.java
@@ -33,6 +33,7 @@ import java.util.concurrent.atomic.AtomicLong;
import java.util.zip.GZIPInputStream;
import static nl.wur.bif.pantools.pangenome.Classification.genome_overview;
+import static nl.wur.bif.pantools.pangenome.create_skip_arrays.create_skip_arrays;
import static nl.wur.bif.pantools.utils.Globals.*;
import static nl.wur.bif.pantools.utils.Utils.complement;
import static nl.wur.bif.pantools.utils.Utils.*;
@@ -1739,6 +1740,7 @@ public class GenomeLayer {
OUTPUT_PATH += "read_mapping/";
new File(WORKING_DIRECTORY + "read_mapping").mkdir();
}
+ create_skip_arrays(false, true);
report_number_of_threads();
int i, j, t, genome;
Node pangenome_node;
@@ -1754,13 +1756,6 @@ public class GenomeLayer {
paired = PATH_TO_THE_SECOND_SRA != null;
if (INTERLEAVED) {
paired = true; // both read pairs are stored in the same file
- }
- if (PATH_TO_THE_GENOME_NUMBERS_FILE == null && target_genome == null) {
- Pantools.logger.error("No genomes were selected to align to.");
- System.exit(1);
- } else if (PATH_TO_THE_GENOME_NUMBERS_FILE != null && target_genome != null) {
- Pantools.logger.error("Include ONLY genome number using a file with --genome-number OR provide the genome number with --reference."); //TODO: this should be impossible
- System.exit(1);
}
FastqReader[] reader = new FastqReader[2];
@@ -1899,66 +1894,20 @@ public class GenomeLayer {
public ArrayList<Integer>[] retrieve_genomes_to_map_against() {
ArrayList<Integer>[] genome_numbers;
genome_numbers = new ArrayList[THREADS];
- int t, number, n = 0;
- String line;
- BufferedReader in;
+ int t, genome_nr;
for (t = 0; t < THREADS; ++t) {
genome_numbers[t] = new ArrayList();
}
- if (PATH_TO_THE_GENOME_NUMBERS_FILE != null) { // --genome-numbers was included
- try {
- in = new BufferedReader(new FileReader(PATH_TO_THE_GENOME_NUMBERS_FILE));
- for (n = 0; (line = in.readLine()) != null; ) {
- line = line.trim();
- if (line.equals(""))
- continue;
- number = Integer.parseInt(line);
- if (number > 0 && number <= GENOME_DB.num_genomes) {
- for (t = 0; t < THREADS; ++t) {
- genome_numbers[t].add(number);
- }
- num_shared_mapping[number] = new AtomicLong(0);
- num_unique_mapping[number] = new AtomicLong(0);
- num_unmapped[number] = new AtomicLong(0);
- ++n;
- } else {
- Pantools.logger.warn("Genome {} is not found in the database.", number);
- }
- }
- in.close();
- } catch (Exception ex) {
- Pantools.logger.warn("Error in reading genome numbers.");
- }
- } else { // --reference was included
- target_genome = target_genome.replace(" ","");
- if (target_genome.endsWith(",")) {
- target_genome = target_genome.replaceFirst(".$",""); // remove last character
- }
- String[] temp_target_array = target_genome.split(",");
- ArrayList<Integer> target_genome_list = new ArrayList<>();
- for (String genome_str : temp_target_array) {
- if (genome_str.contains("-")) {
- String[] genome_array = genome_str.split("-");
- int start = Integer.parseInt(genome_array[0]);
- int end = Integer.parseInt(genome_array[1]);
- for (int i= start; i <= end; i++) {
- target_genome_list.add(i);
- }
- } else {
- target_genome_list.add(Integer.parseInt(genome_str));
- }
- }
- for (int genome_nr : target_genome_list) {
- if (genome_nr > 0 && genome_nr <= GENOME_DB.num_genomes) {
- for (t = 0; t < THREADS; ++t) {
- genome_numbers[t].add(genome_nr);
- }
- num_shared_mapping[genome_nr] = new AtomicLong(0);
- num_unique_mapping[genome_nr] = new AtomicLong(0);
- num_unmapped[genome_nr] = new AtomicLong(0);
- } else {
- Pantools.logger.warn("Genome {} is not found in the database.", genome_nr);
+ for (genome_nr = 1; genome_nr <= GENOME_DB.num_genomes; genome_nr++) {
+ if (!skip_list.contains(genome_nr)) {
+ for (t = 0; t < THREADS; ++t) {
+ genome_numbers[t].add(genome_nr);
}
+ num_shared_mapping[genome_nr] = new AtomicLong(0);
+ num_unique_mapping[genome_nr] = new AtomicLong(0);
+ num_unmapped[genome_nr] = new AtomicLong(0);
+ } else {
+ Pantools.logger.warn("Genome {} is not found in the database.", genome_nr);
}
}
if (ALIGNMENT_MODE < 0) { // when competitive mapping
diff --git a/src/main/java/nl/wur/bif/pantools/pangenome/MultipleSequenceAlignment.java b/src/main/java/nl/wur/bif/pantools/pangenome/MultipleSequenceAlignment.java
index 5362815ef6c7a3474f1086a657c6a428b2fc74d6..1f63e81b25524ca1336b340953f26765e21b9719 100644
--- a/src/main/java/nl/wur/bif/pantools/pangenome/MultipleSequenceAlignment.java
+++ b/src/main/java/nl/wur/bif/pantools/pangenome/MultipleSequenceAlignment.java
@@ -33,6 +33,7 @@ public class MultipleSequenceAlignment {
private final String alignmentTypeCapital; // can be "Protein", "Nucleotide" or "Variants"
private final boolean pavs;
private final List<Long> hmGroups;
+ private final Set<String> excludedGroups;
private ArrayList<Node> hmNodeList;
private ArrayList<String> msaNames;
@@ -53,6 +54,7 @@ public class MultipleSequenceAlignment {
this.alignVariants = alignVariants;
this.pavs = pavs;
this.hmGroups = hmGroups;
+ this.excludedGroups = new HashSet<>();
// set alignmentTypeShort
if (alignProtein) {
@@ -2087,7 +2089,6 @@ public class MultipleSequenceAlignment {
int groupCounter = 0;
ArrayList<String> suitableGroups = new ArrayList<>();
- ArrayList<String> excludedGroups = new ArrayList<>();
for (String groupId : msaNames) {
groupCounter ++;
@@ -2141,7 +2142,7 @@ public class MultipleSequenceAlignment {
* @param excludedGroups alignments not suitable for trimming
* @return true if trimmed, false if not
*/
- private boolean checkIfGroupIsTrimmed(String groupId, Path inPath, ArrayList<String> suitableGroups, ArrayList<String> excludedGroups) {
+ private boolean checkIfGroupIsTrimmed(String groupId, Path inPath, ArrayList<String> suitableGroups, Set<String> excludedGroups) {
boolean pass2 = check_if_file_exists(inPath.resolve("not_trimmed").toString());
if (pass2) {
excludedGroups.add(groupId); // long + str to convert to a string
@@ -2448,6 +2449,16 @@ public class MultipleSequenceAlignment {
return hmNodeList;
}
+
+ /**
+ * Checks if a homology group is excluded based on trimming
+ * @param group homology group node identifier
+ * @return boolean determining if the group is excluded
+ */
+ public boolean isExcluded(long group) {
+ return excludedGroups.contains(Long.toString(group));
+ }
+
/**
* Gives the private variable msaNames
* @return private variable msaNames
diff --git a/src/main/java/nl/wur/bif/pantools/pangenome/Phylogeny.java b/src/main/java/nl/wur/bif/pantools/pangenome/Phylogeny.java
index 59e9dd386421e5d933570d68975dfbead545ff79..7c1730dd060fff775edd37047bd2b6fab3b4e75e 100644
--- a/src/main/java/nl/wur/bif/pantools/pangenome/Phylogeny.java
+++ b/src/main/java/nl/wur/bif/pantools/pangenome/Phylogeny.java
@@ -2635,9 +2635,11 @@ public class Phylogeny {
for (Node hm_node : hmNodeList) {
group_counter++;
Pantools.logger.debug("Reading gene tree: {}/{}.", group_counter, total_groups);
+ if (msa.isExcluded(hm_node.getId())) continue;
gene_tree_file = outDirMsa.resolve(hm_node.getId() + "")
.resolve("output")
.resolve(alignmentTypeShort + trimming + ".newick");
+
gene_tree = read_newick_file(gene_tree_file);
Pantools.logger.debug("Gene tree {} has polytomies: {}", gene_tree_file, hasPolytomies(gene_tree));
diff --git a/src/main/java/nl/wur/bif/pantools/pantools/Pantools.java b/src/main/java/nl/wur/bif/pantools/pantools/Pantools.java
index d55708eeee6cc068766895cf2cde8720e36988e6..fd09b87d4f2bfa052bb8f0009b288d3e4caae389 100644
--- a/src/main/java/nl/wur/bif/pantools/pantools/Pantools.java
+++ b/src/main/java/nl/wur/bif/pantools/pantools/Pantools.java
@@ -52,7 +52,7 @@ import static picocli.CommandLine.ScopeType.INHERIT;
* Implements the main function and declares shared variables. Initializes command line logic.
* Contains the --help and --version commands of the main function.
*
- * @author Roel van Esch, Bioinformatics Group, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Bioinformatics Group, Wageningen University, the Netherlands.
*/
@Command(name = "pantools",
@@ -334,7 +334,7 @@ public class Pantools {
* Custom IHelpSectionRenderer class to use in 'pantools --help'.
* Orders the pantools commands by category.
*
- * @author Roel van Esch
+ * @author Robin van Esch
*/
static class CommandGroupRenderer implements IHelpSectionRenderer {
@@ -432,7 +432,7 @@ public class Pantools {
* Custom IVersionProvider class to use in 'pantools --version'.
* Gets the most up to date version information from git.
*
- * @author Roel van Esch TODO: add authors of getVersionInformation() function (originally in utils/Utils.java)
+ * @author Robin van Esch TODO: add authors of getVersionInformation() function (originally in utils/Utils.java)
*/
protected static class GitVersionProvider implements IVersionProvider {
diff --git a/src/main/java/nl/wur/bif/pantools/utils/BeanUtils.java b/src/main/java/nl/wur/bif/pantools/utils/BeanUtils.java
index 837e81bbe610b7bbde7e0f674af5e1f337f59077..492ea74a456fe04bbce02a9160f5f412f915bb74 100644
--- a/src/main/java/nl/wur/bif/pantools/utils/BeanUtils.java
+++ b/src/main/java/nl/wur/bif/pantools/utils/BeanUtils.java
@@ -16,7 +16,7 @@ import java.util.Set;
/**
* Contains functions for jakarta bean hibernate validation.
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public final class BeanUtils {
diff --git a/src/main/java/nl/wur/bif/pantools/utils/ConsoleUtils.java b/src/main/java/nl/wur/bif/pantools/utils/ConsoleUtils.java
index c8443617257a25cdff6d590a6e20235cdfc1c19d..90f2cd73e3ce59d3254f02b4b80897c1a238b507 100644
--- a/src/main/java/nl/wur/bif/pantools/utils/ConsoleUtils.java
+++ b/src/main/java/nl/wur/bif/pantools/utils/ConsoleUtils.java
@@ -9,7 +9,7 @@ import static java.util.regex.Pattern.CASE_INSENSITIVE;
/**
* Class for console input functions using Scanner
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public final class ConsoleUtils {
diff --git a/src/main/java/nl/wur/bif/pantools/utils/Globals.java b/src/main/java/nl/wur/bif/pantools/utils/Globals.java
index a72d1221c6d39d9db0d667a5d9953838d834b54b..d22e7aafae4c2485d60b22c93938dc04fde6b8ce 100644
--- a/src/main/java/nl/wur/bif/pantools/utils/Globals.java
+++ b/src/main/java/nl/wur/bif/pantools/utils/Globals.java
@@ -77,7 +77,6 @@ public final class Globals {
public static String PATH_TO_THE_PROTEOMES_FILE;
public static String PATH_TO_THE_ANNOTATIONS_FILE;
public static String PATH_TO_THE_REGIONS_FILE;
- public static String PATH_TO_THE_GENOME_NUMBERS_FILE;
public static String RAW_ABUNDANCE_FILE = "";
public static String PATH_TO_THE_FIRST_SRA;
public static String PATH_TO_THE_SECOND_SRA;
diff --git a/src/main/java/nl/wur/bif/pantools/utils/StringUtils.java b/src/main/java/nl/wur/bif/pantools/utils/StringUtils.java
index c1b4a6d64942091347031c9f48b7f193fe0d349c..9627bff837e36e62a5ebe8a7cf20bcd9538cbddf 100644
--- a/src/main/java/nl/wur/bif/pantools/utils/StringUtils.java
+++ b/src/main/java/nl/wur/bif/pantools/utils/StringUtils.java
@@ -9,7 +9,7 @@ import java.util.stream.IntStream;
/**
* Class with string parsing utility functions
*
- * @author Roel van Esch, Wageningen University, the Netherlands.
+ * @author Robin van Esch, Wageningen University, the Netherlands.
*/
public final class StringUtils {
diff --git a/src/main/java/nl/wur/bif/pantools/utils/Utils.java b/src/main/java/nl/wur/bif/pantools/utils/Utils.java
index 4714bec40287a3e683e33271b1684cc606c80d4e..713cf56b25c12cdfab3bb5946e562b46cf294e88 100644
--- a/src/main/java/nl/wur/bif/pantools/utils/Utils.java
+++ b/src/main/java/nl/wur/bif/pantools/utils/Utils.java
@@ -1285,6 +1285,7 @@ public final class Utils {
try (BufferedReader br = new BufferedReader(new FileReader(hmFile.toFile()))) {
String line;
while ((line = br.readLine()) != null) {
+ if (line.startsWith("#")) continue;
String[] hmGroup = line.split(",");
for (String s : hmGroup) {
hmGroups.add(Long.parseLong(s));
diff --git a/src/test/java/nl/wur/bif/pantools/cli/validation/ConstraintsTest.java b/src/test/java/nl/wur/bif/pantools/cli/validation/ConstraintsTest.java
index 9ff2909ac57b586725c336409213962c6f7bc16b..e699ba0b7f05ca676785126747fd9a32d9f3ec90 100644
--- a/src/test/java/nl/wur/bif/pantools/cli/validation/ConstraintsTest.java
+++ b/src/test/java/nl/wur/bif/pantools/cli/validation/ConstraintsTest.java
@@ -21,7 +21,7 @@ import static org.junit.jupiter.api.Assertions.*;
/**
* Unit tests for all constraints within the Constraints class.
*
- * @author Roel van Esch, Bioinformatics group, Wageningen University, the Netherlands
+ * @author Robin van Esch, Bioinformatics group, Wageningen University, the Netherlands
*/
class ConstraintsTest {
diff --git a/src/test/java/nl/wur/bif/pantools/pangenome/ProteomeLayerTest.java b/src/test/java/nl/wur/bif/pantools/pangenome/ProteomeLayerTest.java
index b656f4841aa1629589a56c987ab30f4fc3cfd96b..ec5d0bc67d5d25a3fd58f6cfbb9f1996dcd1b9c7 100644
--- a/src/test/java/nl/wur/bif/pantools/pangenome/ProteomeLayerTest.java
+++ b/src/test/java/nl/wur/bif/pantools/pangenome/ProteomeLayerTest.java
@@ -28,7 +28,7 @@ import static org.junit.jupiter.api.Assertions.*;
/**
* Unit tests for all functions within the ProteomeLayer class.
*
- * @author Roel van Esch, Bioinformatics group, Wageningen University, the Netherlands
+ * @author Robin van Esch, Bioinformatics group, Wageningen University, the Netherlands
*/
public class ProteomeLayerTest {
diff --git a/tests/Snakefile b/tests/Snakefile
index e5a1b08f5c67389d0418111db438960141f945fc..28a202f071406001960eca0cca00d5242d2ea6f0 100644
--- a/tests/Snakefile
+++ b/tests/Snakefile
@@ -9,12 +9,12 @@ It will check:
To execute, specify a configuration file with the data set, e.g.:
-snakemake --cores 1 --configfile yeast-1.yaml
+snakemake --cores 1 --configfile yeast.yaml
The pipeline will package the local version into a .jar, as well as a reference
-version (a git revision) specified in shared.yaml. It will download the dataset
+version (a git revision) specified in shared.yaml. It will use the dataset
referenced by the configuration file passed in by the user with --configfile
-(e.g. yeast-1.yaml, yeast-4.yaml), and download and untar the dataset.
+(e.g. yeast.yaml, etc.).
A pangenome will be built with `build_pangenome` for the local and reference
versions each. Pangenomes will be exported with `export_pangenome`, the
@@ -125,68 +125,42 @@ rule export_pangenome:
--sequence-node-anchors-file {output.sequence_node_anchors}
"""
-rule download_dataset:
- """Download the data set with the provided name."""
- # TODO: URL should be changed once data is uploaded
- # TODO: md5sum check?
- # TODO: split genomes.txt generation from downloading data set
- # TODO: standardize data set formats, file names, etc.
+rule list_genomes:
+ """Lists the input genomes."""
+ input:
+ genomes=config["dataset"]["genomes"],
output:
genome_list="input/data/{dataset_name}/genomes/genomes.txt",
- fastq1="input/data/{dataset_name}/reads/1.fastq",
- fastq2="input/data/{dataset_name}/reads/2.fastq"
- params:
- url=config["dataset"]["url"]
shell:
"""
- mkdir -p "input/data/{wildcards.dataset_name}/"
-
- # TODO: for some reason a streaming untar does not seem to work (anymore)
- curl -o dataset.tar --silent {params.url}
- tar xf dataset.tar -C input/data/{wildcards.dataset_name}/
- rm dataset.tar
-
- find input/data/{wildcards.dataset_name}/genomes/ -iname '*.fasta' | sort > {output.genome_list}
-
- mv input/data/{wildcards.dataset_name}/reads/*_1.fastq {output.fastq1}
- mv input/data/{wildcards.dataset_name}/reads/*_2.fastq {output.fastq2}
+ ls {input.genomes} | sort > {output.genome_list}
"""
rule map_reads:
"""Map reads against a a database, outputs a SAM file per genome."""
input:
- fastq1="input/data/{dataset_name}/reads/1.fastq",
- fastq2="input/data/{dataset_name}/reads/2.fastq",
- genome_list="input/data/{dataset_name}/genomes/genomes.txt",
+ fastq1=config["dataset"]["reads"]["fastq1"],
+ fastq2=config["dataset"]["reads"]["fastq2"],
database="output/databases/{dataset_name}/{version}/",
jar="jars/pantools/target/pantools-{version}.jar"
output:
sam_files=expand(
"output/alignments/{{dataset_name}}/{{version}}/sams/pantools_{genome}.sam",
genome=range(1, len(config["dataset"]["genomes"]) + 1)
- ),
- genome_numbers_file="output/alignments/{dataset_name}/{version}/genome-numbers.txt"
- params:
- num_genomes=len(config["dataset"]["genomes"])
+ )
shell:
"""
- mkdir -p $(dirname {output.genome_numbers_file})
-
- seq 1 {params.num_genomes} > {output.genome_numbers_file}
-
{java} \
-jar {input.jar} \
map \
--threads 1 \
{input.database} \
- {output.genome_numbers_file} \
{input.fastq1} \
{input.fastq2} \
--gap-open -20 \
--gap-extension -3 \
--out-format SAM \
- --output $(dirname {output.sam_files[0]}) \
- # --reference 1-$(wc -l <{input.genome_list} | sed 's/ //g')
+ --output $(dirname {output.sam_files[0]})
"""
rule strip_sam_pg_header_line:
diff --git a/tests/shared.yaml b/tests/shared.yaml
index fe62639baa6e31abe0f4b13d19cd24465b95bc5e..c32a5ae8adbd6e3e992469c1f3c948c1f393f7e3 100644
--- a/tests/shared.yaml
+++ b/tests/shared.yaml
@@ -2,7 +2,7 @@
tools:
pantools:
repository_url: https://git.wur.nl/bioinformatics/pantools.git
- reference_version: 8da5bd05
+ reference_version: 0c4a3485
java_arguments:
- -Xmx8g
maven:
diff --git a/tests/yeast-1.yaml b/tests/yeast-1.yaml
deleted file mode 100644
index 0eebe780084daa5fc2c324df35a7a85cef609534..0000000000000000000000000000000000000000
--- a/tests/yeast-1.yaml
+++ /dev/null
@@ -1,8 +0,0 @@
-dataset:
- name: "yeast-1"
- url: "https://www.bioinformatics.nl/pangenomics/data/ci/yeast-1.tar.gz"
- genomes:
- - GCF_000146045.2_R64_genomic.fasta
- reads:
- fastq1: 1_MSv3_1.fastq
- fastq2: 1_MSv3_2.fastq
diff --git a/tests/yeast-4.yaml b/tests/yeast-4.yaml
deleted file mode 100644
index 815abcef0a45e14b20c26728fc2f25770bb66c11..0000000000000000000000000000000000000000
--- a/tests/yeast-4.yaml
+++ /dev/null
@@ -1,11 +0,0 @@
-dataset:
- name: "yeast-4"
- url: "https://www.bioinformatics.nl/pangenomics/data/ci/yeast-4.tar.gz"
- genomes:
- - GCA_000167035.1_ASM16703v1_genomic.fasta
- - GCA_000256765.1_Saccharomyces_kudriavzevii_strain_FM1066_v1.0_genomic.fasta
- - GCF_000146045.2_R64_genomic.fasta
- - GCF_001298625.1_SEUB3.0_genomic.fasta
- reads:
- fastq1: 1_MSv3_1.fastq
- fastq2: 1_MSv3_2.fastq
diff --git a/tests/yeast.yaml b/tests/yeast.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..1dedfdae4341334b7bda377065caa9c0fdb9b8eb
--- /dev/null
+++ b/tests/yeast.yaml
@@ -0,0 +1,11 @@
+dataset:
+ name: "yeast"
+ genomes:
+ - yeast_tiny_pangenome/genomes/273614_chrI.fna
+ - yeast_tiny_pangenome/genomes/BY_chrI.fna
+ - yeast_tiny_pangenome/genomes/CBS2888_chrI.fna
+ - yeast_tiny_pangenome/genomes/S288C_chrI.fna
+ - yeast_tiny_pangenome/genomes/Y10_chrI.fna
+ reads:
+ fastq1: yeast_tiny_pangenome/reads/SRR800844_chrI_1.fq.gz
+ fastq2: yeast_tiny_pangenome/reads/SRR800844_chrI_2.fq.gz