Proteins only homology evidence

Enables use of protein sequences as evidence for homology module
EI-CoreBioinformatics · Sep 4, 2021 · 1b55933 · 1b55933
1 parent 339877f
commit 1b55933
Show file tree

Hide file tree

Showing 5 changed files with 973 additions and 4 deletions.
diff --git a/annotation/__main__.py b/annotation/__main__.py
@@ -377,8 +377,9 @@ def parse_arguments():
     homology_ap.add_argument("--annotations_csv", type=argparse.FileType('r'),
                              help="CSV file with reference annotations to extract proteins/cdnas for spliced alignments"
                                   " in csv format. The CSV fields are as follows genome_fasta,annotation_gff  "
-                                  "e.g Athaliana.fa,Athaliana.gff",
-                             required=True)
+                                  "e.g Athaliana.fa,Athaliana.gff")
+    homology_ap.add_argument("--protein_sequences", type=str, nargs='*',
+                             help="List of files containing protein sequences to use as evidence")
     homology_ap.add_argument("--annotation_filters",
                              choices=['all', 'none', 'exon_len', 'intron_len', 'internal_stop', 'aa_len',
                                       'splicing'],

diff --git a/annotation/homology.py b/annotation/homology.py
@@ -22,6 +22,9 @@ def combine_arguments_homology(cli_arguments):
     cromwell_inputs["ei_homology.mikado_scoring"] = cli_arguments.mikado_scoring.name
     cromwell_inputs["ei_homology.output_prefix"] = cli_arguments.output_prefix
 
+    if cli_arguments.protein_sequences:
+        cromwell_inputs["ei_homology.protein_sequence_files"] = cli_arguments.protein_sequences
+
     # Optional extra parameters
     if cli_arguments.pick_extra_config:
         cromwell_inputs['ei_homology.MikadoPick.extra_config'] = cli_arguments.pick_extra_config.name

diff --git a/tests/integration/homology/annotations.csv b/tests/integration/homology/annotations.csv
@@ -1,5 +1,4 @@
 tests/integration/homology/inputs/BrapaFPsc_Chr3_1065466-1464870.fa,tests/integration/homology/inputs/BrapaFPsc_Chr3_1065466-1464870.gff
-tests/integration/homology/inputs/Cgrandiflora_Chr3_1065466-1464870.fa,tests/integration/homology/inputs/Cgrandiflora_Chr3_1065466-1464870.gff
 tests/integration/homology/inputs/Crubella_Chr3_1065466-1464870.fa,tests/integration/homology/inputs/Crubella_Chr3_1065466-1464870.gff
 tests/integration/homology/inputs/Ptrichocarpa_Chr3_1065466-1464870.fa,tests/integration/homology/inputs/Ptrichocarpa_Chr3_1065466-1464870.gff
 tests/integration/homology/inputs/Vvinifera_Chr3_1065466-1464870.fa,tests/integration/homology/inputs/Vvinifera_Chr3_1065466-1464870.gff