Merge branch 'master' into 'lcs'

(cherry picked some changes. Will make further polishing)
# Conflicts:
#   Installation Guide.md
#   bigscape.py

ArrowerSVG.py

@@ -8,6 +8,12 @@
 #                heavily modified by Jorge Navarro 2016              #
 ######################################################################
 
+# Makes sure the script can be used with Python 2 as well as Python 3.
+from __future__ import print_function, division
+from sys import version_info
+if version_info[0]==2:
+    range = xrange
+
 import os
 import sys
 import argparse
@@ -169,15 +175,10 @@ def draw_arrow(additional_tabs, X, Y, L, l, H, h, strand, color, color_contour,
     else:
         color_contour = [50, 50, 50]
         
-    arrow += additional_tabs + "\t\t<polygon "
-    arrow += "class=\"" + gid + "\" "
-    arrow += "points=\"" + " ".join(points_coords) + "\" "
-    arrow += "fill=\"rgb(" + ",".join(map(str,color)) +")\" "
-    arrow += "fill-opacity=\"1.0\" "
-    arrow += "stroke=\"rgb(" + ",".join(map(str,color_contour)) + ")\" "
-    arrow += "stroke-width=\"" + str(gene_contour_thickness) + "\" " 
-    arrow += category + " />\n"
-
+    arrow += "{}\t\t<polygon class=\"{}\" ".format(additional_tabs, gid)
+    arrow += "points=\"{}\" fill=\"rgb({})\" ".format(" ".join(points_coords), ",".join([str(val) for val in color]))
+    arrow += "fill-opacity=\"1.0\" stroke=\"rgb({})\" ".format(",".join([str(val) for val in color_contour]))
+    arrow += "stroke-width=\"{}\" {} />\n".format(str(gene_contour_thickness), category)
     
     # paint domains. Domains on the tip of the arrow should not have corners sticking
     #  out of them
@@ -193,8 +194,7 @@ def draw_arrow(additional_tabs, X, Y, L, l, H, h, strand, color, color_contour,
         dccolour = domain[6]
         
         arrow += additional_tabs + "\t\t<g>\n"
-        arrow += additional_tabs + "\t\t\t<title>" + dname + " (" + dacc + ")\n\"" + ddesc + "\"</title>\n"
-        
+        arrow += "{}\t\t\t<title>{} ({})\n\"{}\"</title>\n".format(additional_tabs, dname, dacc, ddesc)
         
         if strand == "+":
             # calculate how far from head_start we (the horizontal guide at y=Y+internal_domain_margin)
@@ -211,16 +211,11 @@ def draw_arrow(additional_tabs, X, Y, L, l, H, h, strand, color, color_contour,
             x_margin_offset = internal_domain_margin/sin(pi - atan2(h+H/2.0,-head_length))
 
             if (dX + dL) < head_start + collision_x - x_margin_offset:
-                arrow += additional_tabs + "\t\t\t<rect class=\"" + dacc + "\" "
-                arrow += "x=\"" + str(X+dX) + "\" "
-                arrow += "y=\"" + str(Y + internal_domain_margin) + "\" "
-                arrow += "stroke-linejoin=\"round\" "
-                arrow += "width=\"" + str(dL) + "\" "
-                arrow += "height=\"" + str(dH) + "\" "
-                arrow += "fill=\"rgb(" + ",".join(map(str,dcolor)) + ")\" "
-                arrow += "stroke=\"rgb(" + ",".join(map(str,dccolour)) + ")\" "
-                arrow += "stroke-width=\"" + str(domain_contour_thickness) + "\" "
-                arrow += "opacity=\"0.75\" />\n"
+                arrow += "{}\t\t\t<rect class=\"{}\" x=\"{}\" ".format(additional_tabs, dacc, str(X+dX))
+                arrow += "y=\"{}\" stroke-linejoin=\"round\" ".format(str(Y + internal_domain_margin))
+                arrow += "width=\"{}\" height=\"{}\" ".format(str(dL), str(dH))
+                arrow += "fill=\"rgb({})\" stroke=\"rgb({})\" ".format(",".join([str(val) for val in dcolor]), ",".join([str(val) for val in dccolour]))
+                arrow += "stroke-width=\"{}\" opacity=\"0.75\" />\n".format(str(domain_contour_thickness))
             else:
                 del points[:]
                 
@@ -263,15 +258,12 @@ def draw_arrow(additional_tabs, X, Y, L, l, H, h, strand, color, color_contour,
                 for point in points:
                     points_coords.append(str(int(point[0])) + "," + str(int(point[1])))
                     
-                arrow += additional_tabs + "\t\t\t<polygon class=\"" + dacc + "\" "
-                arrow += "points=\"" + " ".join(points_coords) + "\" "
-                arrow += "stroke-linejoin=\"round\" "
-                arrow += "width=\"" + str(dL) + "\" "
-                arrow += "height=\"" + str(dH) + "\" "
-                arrow += "fill=\"rgb(" + ",".join(map(str,dcolor)) + ")\" "
-                arrow += "stroke=\"rgb(" + ",".join(map(str,dccolour)) + ")\" "
-                arrow += "stroke-width=\"" + str(domain_contour_thickness) + "\" "
-                arrow += "opacity=\"0.75\" />\n"    
+                arrow += "{}\t\t\t<polygon class=\"{}\" ".format(additional_tabs, dacc)
+                arrow += "points=\"{}\" stroke-linejoin=\"round\" ".format(" ".join(points_coords))
+                arrow += "width=\"{}\" height=\"{}\" ".format(str(dL), str(dH))
+                arrow += "fill=\"rgb({})\" ".format(",".join([str(val) for val in dcolor]))
+                arrow += "stroke=\"rgb({})\" ".format(",".join([str(val) for val in dccolour]))
+                arrow += "stroke-width=\"{}\" opacity=\"0.75\" />\n".format(str(domain_contour_thickness))
             
         # now check other direction
         else:
@@ -286,16 +278,12 @@ def draw_arrow(additional_tabs, X, Y, L, l, H, h, strand, color, color_contour,
             
             # nice, blocky domains
             if dX > collision_x + x_margin_offset:
-                arrow += additional_tabs + "\t\t\t<rect class=\"" + dacc + "\" "
-                arrow += "x=\"" + str(X+dX) + "\" "
-                arrow += "y=\"" + str(Y + internal_domain_margin) + "\" "
-                arrow += "stroke-linejoin=\"round\" "
-                arrow += "width=\"" + str(dL) + "\" "
-                arrow += "height=\"" + str(dH) + "\" "
-                arrow += "fill=\"rgb(" + ",".join(map(str,dcolor)) + ")\" "
-                arrow += "stroke=\"rgb(" + ",".join(map(str,dccolour)) + ")\" "
-                arrow += "stroke-width=\"" + str(domain_contour_thickness) + "\" "
-                arrow += "opacity=\"0.75\" />\n"
+                arrow += "{}\t\t\t<rect class=\"{}\" ".format(additional_tabs, dacc)
+                arrow += "x=\"{}\" y=\"{}\" ".format(str(X+dX), str(Y + internal_domain_margin))
+                arrow += "stroke-linejoin=\"round\" width=\"{}\" height=\"{}\" ".format(str(dL), str(dH))
+                arrow += "fill=\"rgb({})\" ".format(",".join([str(val) for val in dcolor]))
+                arrow += "stroke=\"rgb({})\" ".format(",".join([str(val) for val in dccolour]))
+                arrow += "stroke-width=\"{}\" opacity=\"0.75\" />\n".format(str(domain_contour_thickness))
             else:
                 del points[:]
                 
@@ -329,15 +317,12 @@ def draw_arrow(additional_tabs, X, Y, L, l, H, h, strand, color, color_contour,
                 for point in points:
                     points_coords.append(str(int(point[0])) + "," + str(int(point[1])))
                     
-                arrow += additional_tabs + "\t\t\t<polygon class=\"" + dacc + "\" "
-                arrow += "points=\"" + " ".join(points_coords) + "\" "
-                arrow += "stroke-linejoin=\"round\" "
-                arrow += "width=\"" + str(dL) + "\" "
-                arrow += "height=\"" + str(dH) + "\" "
-                arrow += "fill=\"rgb(" + ",".join(map(str,dcolor)) + ")\" "
-                arrow += "stroke=\"rgb(" + ",".join(map(str,dccolour)) + ")\" "
-                arrow += "stroke-width=\"" + str(domain_contour_thickness) + "\" "
-                arrow += "opacity=\"0.75\" />\n"
+                arrow += "{}\t\t\t<polygon class=\"{}\" ".format(additional_tabs, dacc)
+                arrow += "points=\"{}\" stroke-linejoin=\"round\" ".format(" ".join(points_coords))
+                arrow += "width=\"{}\" height=\"{}\" ".format(str(dL), str(dH))
+                arrow += "fill=\"rgb({})\" ".format(",".join([str(val) for val in dcolor]))
+                arrow += "stroke=\"rgb({})\" ".format(",".join([str(val) for val in dccolour]))
+                arrow += "stroke-width=\"{}\" opacity=\"0.75\" />\n".format(str(domain_contour_thickness))
         
         arrow += additional_tabs + "\t\t</g>\n"
     
@@ -351,7 +336,7 @@ def draw_line(X,Y,L):
     Draw a line below genes
     """
     
-    line = "<line x1=\"" + str(X) + "\" y1=\"" + str(Y) + "\" x2=\"" + str(X+L) + "\" y2=\"" + str(Y) + "\" style=\"stroke:rgb(50,50,50); stroke-width:" + str(stripe_thickness) + " \"/>\n"
+    line = "<line x1=\"{}\" y1=\"{}\" x2=\"{}\" y2=\"{}\" style=\"stroke:rgb(50,50,50); stroke-width:{} \"/>\n".format(str(X), str(Y), str(X+L), str(Y), str(stripe_thickness))
     
     return line
 
@@ -426,7 +411,7 @@ def SVG(write_html, outputfile, GenBankFile, pfdFile, use_pfd, color_genes, colo
         header = "\t\t<div title=\"" + GenBankFile[:-4] + "\">\n"
         additional_tabs = "\t\t\t"
         
-        header += additional_tabs + "<svg width=\"" + str(max_width + 2*(mX)) + "\" height=\"" + str(loci*(2*h + H + 2*mY)) + "\">\n"
+        header += "{}<svg width=\"{}\" height=\"{}\">\n".format(additional_tabs, str(max_width + 2*(mX)), str(loci*(2*h + H + 2*mY)))
 
         addY = loci*(2*h + H + 2*mY)
     else:
@@ -654,7 +639,7 @@ def SVG(write_html, outputfile, GenBankFile, pfdFile, use_pfd, color_genes, colo
             
         with open(gene_color_file, "a") as color_genes_handle:
             for new_names in new_color_genes:
-                color_genes_handle.write(new_names + "\t" + ",".join(map(str,new_color_genes[new_names])) + "\n")
+                color_genes_handle.write(new_names + "\t" + ",".join([str(ncg) for ncg in new_color_genes[new_names]]) + "\n")
     
     if len(new_color_domains) > 0:
         if len(new_color_domains) < 10:
@@ -664,7 +649,7 @@ def SVG(write_html, outputfile, GenBankFile, pfdFile, use_pfd, color_genes, colo
             
         with open(domains_color_file, "a") as color_domains_handle:
             for new_names in new_color_domains:
-                color_domains_handle.write(new_names + "\t" + ",".join(map(str,new_color_domains[new_names])) + "\n")
+                color_domains_handle.write(new_names + "\t" + ",".join([str(ncdom) for ncdom in new_color_domains[new_names]]) + "\n")
 
     
     mode = "a" if write_html == True else "w"

Installation Guide.md

@@ -31,6 +31,12 @@ installation using Miniconda
 > conda install -c bioconda hmmer biopython mafft
 ```
 
-* The Affinity Propagation algorithm used currently is [pySAPC](https://pypi.python.org/pypi/pysapc/1.1.0). If you are using a Mac: `conda install -c https://conda.anaconda.org/bioinfocao pysapc`, otherwise: `pip install pysapc`
+* The Affinity Propagation algorithm used currently is [pySAPC](https://pypi.python.org/pypi/pysapc/1.1.0). Install cython first:
 
-(pysapc also installs pandas, cython, pytz, six and python-dateutil)
+```
+> conda install cython
+```
+
+then, if you are using a Mac: `conda install -c https://conda.anaconda.org/bioinfocao pysapc`, otherwise: `pip install pysapc`
+
+(pysapc also installs pandas, pytz, six and python-dateutil)

domains_color_file.tsv

@@ -4015,3 +4015,4 @@ PF05938	207,71,220
 PF08082	78,184,113
 PF07725	90,147,197
 PF01582	219,220,68
+PF07678	207,214,67

functions.py

@@ -16,6 +16,11 @@ Functions used by bigscape.py
 # License: GNU Affero General Public License v3 or later
 # A copy of GNU AGPL v3 should have been included in this software package in LICENSE.txt.
 """
+# Makes sure the script can be used with Python 2 as well as Python 3.
+from __future__ import print_function
+from sys import version_info
+if version_info[0]==2:
+    range = xrange
 
 import os
 import subprocess
@@ -62,9 +67,9 @@ def get_anchor_domains(filename):
                     domains.add(line.strip().split("\t")[0].split(".")[0])
         return domains
     except IOError:
-        print "You have not provided the anchor_domains.txt file."
-        print "if you want to make use of the anchor domains in the DSS distance metric,\
-        make a file that contains a Pfam domain on each line."
+        print("You have not provided the anchor_domains.txt file.")
+        print("if you want to make use of the anchor domains in the DSS distance metric, \
+make a file that contains a Pfam domain on each line.")
         return set()
         
 
@@ -157,15 +162,6 @@ def write_pfd(pfd_handle, matrix):
     pfd_handle.close() 
     
 
-def get_domains(filename):
-    handle = open(filename, 'r')
-    domains = []
-    for line in handle:
-        if line[0] != "#":
-            domains.append(filter(None, line.split(" "))[1])
-            
-    return domains
-
 
 def no_overlap(locA1, locA2, locB1, locB2):
     """Return True if there is no overlap between two regions"""
@@ -229,11 +225,9 @@ def save_domain_seqs(filtered_matrix, fasta_dict, domains_folder, outputbase):
         header = row[-1].strip()
         seq = fasta_dict[header] #access the sequence by using the header
         
-
         domain_file = open(os.path.join(domains_folder, domain + ".fasta"), 'a') #append to existing file
-        domain_file.write(">" + header + ":" + row[3] + ":" + row[4] \
-        + "\n" + seq[int(row[3]):int(row[4])] + "\n") #only use the range of the pfam domain within the sequence
-            
+        domain_file.write(">{}:{}:{}\n{}\n".format(header, row[3], row[4],
+            seq[int(row[3]):int(row[4])])) #only use the range of the pfam domain within the sequence
         domain_file.close()
 
 
@@ -430,7 +424,7 @@ def domtable_parser(gbk, dom_file):
     else:
         for line in dom_handle:        
             if line[0] != "#":
-                splitline = filter(None, line.split(" "))
+                splitline = line.split()
                 pfd_row = []
                 pfd_row.append(gbk)         #add clustername or gbk filename
                 
@@ -440,7 +434,7 @@ def domtable_parser(gbk, dom_file):
                 try:
                     pfd_row.append(header_list[header_list.index("gid")+1]) #add gene ID if known
                 except ValueError:
-                    print "No gene ID in ", gbk
+                    print("No gene ID in " + gbk)
                     pfd_row.append('')
                     
                 pfd_row.append(splitline[19])#first coordinate, env coord from

gene_color_file.tsv

@@ -12759,3 +12759,4 @@ Araha.8081s0001	211,252,249
 Araha.24309s0001	182,250,217
 Araha.24309s0002	221,251,191
 Araha.24309s0003	242,205,251
+eno	192,245,254