Merge branch 'devel'

bin/build_baited_bloom_filter

@@ -2,13 +2,15 @@
 
 import argparse
 import logging
-from exfi import __version__
-from exfi.build_baited_bloom_filter import build_baited_bloom_filter
 import sys
 from os.path import isfile, exists, dirname, abspath
 from os import \
     makedirs, \
     remove
+from shutil import which
+
+from exfi import __version__
+from exfi.build_baited_bloom_filter import build_baited_bloom_filter
 
 parser = argparse.ArgumentParser(
     usage='build_baited_bloom_filter '
@@ -122,21 +124,25 @@ if __name__ == '__main__':
 
     # Set up logger
     logger = logging.getLogger()
-    logging.basicConfig(format='%(asctime)s\t%(module)s\t%(message)s', level=logging.ERROR)
+    logging.basicConfig(
+        format='%(asctime)s\t%(module)s\t%(message)s',
+        level=logging.ERROR
+    )
     if args["verbose"]:
         logger.setLevel(logging.INFO)
     if args["debug"]:
         logger.setLevel(logging.DEBUG)
 
 
     # Check inputs
+    logging.info('Checking input parameters')
     assert args["kmer"] >= 1, 'ERROR: incorrect kmer size'
     # assert bloom_size
     assert args["levels"] >= 1, 'ERROR: incorrect number of levels'
     assert args["threads"] >= 1, 'ERROR: incorrect number of threads'
 
     # Check if programs are in path
-    from shutil import which
+    logging.info('Checking if biobloom* and abyss-bloom are in $PATH')
     assert which('biobloommaker') is not None, 'ERROR: biobloommaker not in PATH'
     assert which('biobloomcategorizer') is not None, 'ERROR: biobloomcategorizer not in PATH'
     assert which('abyss-bloom') is not None, 'ERROR: abyss-bloom not in PATH'
@@ -147,6 +153,7 @@ if __name__ == '__main__':
         makedirs(output_dir)
 
     # Run the program
+    logging.info('Running build_baited_bloom_filter')
     build_baited_bloom_filter(args)
 
     logging.info("Done!")

bin/build_splice_graph

@@ -11,33 +11,18 @@ from os.path import \
 from Bio import SeqIO
 
 from exfi import __version__
-
-from exfi.find_exons import \
-    _get_fasta, \
-    _find_exons_pipeline
-
-from exfi.build_splice_graph_dict import \
-    build_splice_graph_dict
-
-from exfi.correct import \
-    correct_splice_graph_dict
-
-from exfi.polish import \
-    polish_splice_graph_dict
-
-from exfi.io.fasta_to_dict import \
-    fasta_to_dict
-
-from exfi.io.splice_graph_dict_to_gfa1 import \
-    splice_graph_dict_to_gfa1
-
-from exfi.collapse import \
-    collapse_splice_graph_dict
+from exfi.find_exons import find_exons
+from exfi.io.bed import bed3_to_bed4
+from exfi.io.fasta_to_dict import fasta_to_dict
+from exfi.polish import polish_bed4
+from exfi.correct import correct_bed4
+from exfi.io.bed4_to_gfa1 import bed4_to_gfa1
 
 
 
 parser = argparse.ArgumentParser(
-    usage='build_splicegraph -i transcriptome.fa -b bloom_filter.bf -k 30 -o exome.gfa',
+    usage='build_splicegraph -i transcriptome.fa -b bloom_filter.bf -k 30 '
+          '-o exome.gfa',
     description='Store the predicted exome in GFA format',
     epilog='Jorge Langa. Send issues and pull requests to github.com/jlanga/'
            'exfi',
@@ -137,7 +122,8 @@ parser.add_argument(
 parser.add_argument(
     '--correct',
     '-C',
-    help='Correct splice graph by using sealer between exons that seem nearby [False]',
+    help='Correct splice graph by using sealer between exons that seem nearby '
+        '[False]',
     dest='correct',
     action="store_true"
 )
@@ -190,33 +176,22 @@ if __name__ == "__main__":
         logger.setLevel(logging.DEBUG)
 
     # Get predicted exons in bed format
-    positive_exons_bed = _find_exons_pipeline(args)
-
-    # Build splice graph
-    splice_graph_dict = build_splice_graph_dict(positive_exons_bed, args)
+    bed4 = bed3_to_bed4(find_exons(args))
 
     # Transcriptome_dict
     transcriptome_dict = fasta_to_dict(args["fasta"])
 
     if args["polish"]:
-        splice_graph_dict = polish_splice_graph_dict(splice_graph_dict, transcriptome_dict, args)
+        bed4 = polish_bed4(bed4=bed4, transcriptome_dict=transcriptome_dict)
 
     if args["correct"]:
-        splice_graph_dict = correct_splice_graph_dict(splice_graph_dict, args)
-
-    if args["collapse"]:
-        splice_graph_dict = {
-            "collapsed": collapse_splice_graph_dict(
-                splice_graph_dict=splice_graph_dict,
-                transcriptome_dict=transcriptome_dict
-            )
-        }
+        bed4 = correct_bed4(
+            bed4=bed4, transcriptome_dict=transcriptome_dict, args=args
+        )
 
     # Write to GFA1
-    splice_graph_dict_to_gfa1(
-        splice_graph_dict=splice_graph_dict,
-        transcriptome_dict=transcriptome_dict,
-        filename=args["gfa1"]
+    bed4_to_gfa1(
+        gfa1_fn=args["gfa1"], bed4=bed4, transcriptome_dict=transcriptome_dict
     )
 
     logging.info("Done!")

bin/gfa1_to_exons

@@ -10,7 +10,7 @@ from os.path import \
 
 from exfi import __version__
 
-from exfi.io.gfa1_to_exons import \
+from exfi.io.gfa1_to_fasta import \
     gfa1_to_exons
 
 parser = argparse.ArgumentParser(
@@ -34,7 +34,8 @@ parser.add_argument(
     '-i',
     type=str,
     required=True,
-    help='Input splice graph in GFA1 format (the results from build_splicegraph)',
+    help='Input splice graph in GFA1 format (the results from '
+         'build_splicegraph)',
     dest='gfa1',
     metavar='FILE'
 )
@@ -107,8 +108,8 @@ if __name__ == "__main__":
         masking = "hard"
 
     gfa1_to_exons(
-        gfa_in_fn=args["gfa1"],
-        fasta_out_fn=args["fasta"],
+        fasta_out=args["fasta"],
+        gfa1_in=args["gfa1"],
         masking=masking
     )
 

bin/gfa1_to_gapped_transcripts

@@ -9,7 +9,7 @@ from os.path import \
     abspath
 
 from exfi import __version__
-from exfi.io.gfa1_to_gapped_transcripts import \
+from exfi.io.gfa1_to_fasta import \
     gfa1_to_gapped_transcripts
 
 
@@ -121,9 +121,9 @@ if __name__ == "__main__":
         masking = "hard"
 
     gfa1_to_gapped_transcripts(
-        gfa_in=args["gfa1"],
         fasta_out=args["fasta"],
-        number_of_ns=args["number_of_ns"],
+        gfa1_in=args["gfa1"],
+        gap_size=args["number_of_ns"],
         masking=masking
     )
 

exfi/__init__.py

@@ -3,4 +3,4 @@
 filters.
 """
 
-__version__ = '1.4.12'
+__version__ = '1.4.13'