Fast5 mode of get_one_barcode.py now copies files directly instead of outputting the filenames

tools/get_one_barcode.py

@@ -5,22 +5,23 @@ from pathlib import Path
 import pandas as pd
 from Bio import SeqIO
 import argparse
+import shutil
 
 # Never truncate strings:
 pd.set_option('display.max_colwidth', None)
 
 
-def get_barcode_fast5_reads(barcode_id, sequencing_summary_path, sequencing_dir,
-                            out_dir, worker_count):
-    """Get list of all fast5 files that belong to a certain barcode
+def copy_barcoded_fast5_reads(barcode_id, sequencing_summary_path, sequencing_dir,
+                              out_dir, worker_count):
+    """Copy all fast5 files that belong to a certain barcode to out_dir
 
     :param barcode_id: barcode to extract, example 'barcode12'
     :type barcode_id: str
     :param sequencing_summary_path: path to sequencing_summary.txt
     :param sequencing_dir: path to directory containing fast5 reads
-    :param out_dir: where to save output filenames and what to call the file
+    :param out_dir: where to save fast5 files
     :param worker_count: how many processes to run in parallel
-    :return: list of fast 5 filenames
+    :return: list of fast5 filenames
     """
     # Index all files
     all_files = os.walk(sequencing_dir)
@@ -38,9 +39,9 @@ def get_barcode_fast5_reads(barcode_id, sequencing_summary_path, sequencing_dir,
     # Merge list of lists into one big list
     files_of_read = itertools.chain.from_iterable(files_of_read)
 
-    print('Saving to file')
-    series = pd.Series(files_of_read)
-    series.to_csv(out_dir, index=False, header=False)
+    os.makedirs(out_dir, exist_ok=True)
+    for file in files_of_read:
+        shutil.copy(file, out_dir)
 
     return files_of_read
 
@@ -159,8 +160,8 @@ def main():
         get_barcode_fastq_reads(barcode, seq_summary_path, read_path,
                                 out_dir, worker_count)
     elif args.input_type == 'fast5':
-        get_barcode_fast5_reads(barcode, seq_summary_path, read_path,
-                                out_dir, worker_count)
+        copy_barcoded_fast5_reads(barcode, seq_summary_path, read_path,
+                                  out_dir, worker_count)
 
 
 if __name__ == '__main__':