Merge pull request #93 from MicrobialDarkMatter/bugfix/pileup-in-binc…

…onsensus-invalid-argument

Argparser argument fixes

.gitignore

@@ -1,4 +1,5 @@
 data
+tests/cli/
 nanomotif/flamegraph*
 austin
 results

nanomotif/_version.py

@@ -1 +1 @@
-__version__ = "0.5.2"
+__version__ = "0.5.3"

nanomotif/argparser.py

@@ -47,6 +47,7 @@ def create_parser():
     )
     parser_score_motifs.add_argument("motifs", type=str, help="path to the motifs file.")
     parser_score_motifs.add_argument("--save-motif-positions", action="store_true", help="save motif positions in the output folder")
+    parser_score_motifs.add_argument("--threshold_valid_coverage", type=int, default=5, help="minimum valid base coverage for a position to be considered. Default: %(default)s")
 
     ###########################################################################
     # Bin consensus
@@ -60,6 +61,7 @@ def create_parser():
     parser_bin_consensus.add_argument("motifs", type=str, help="path to the motifs file.")
     parser_bin_consensus.add_argument("bins", type=str, help="tsv file specifying which bin contigs belong.")
     parser_bin_consensus.add_argument("motifs_scored", metavar="motifs-scored", type=str, help="path to the motif-scored file.")
+    parser_bin_consensus.add_argument("--threshold_valid_coverage", type=int, default=5, help="minimum valid base coverage for a position to be considered. Default: %(default)s")
 
     ###########################################################################
     # Complete workflow

nanomotif/datasets/geobacillus-plasmids.motifs-scored.tsv

@@ -0,0 +1,9 @@
+contig	mod_type	motif	mod_position	n_mod	n_nomod	motif_type	motif_complement	mod_position_complement	n_mod_complement	n_nomod_complement
+contig_2	a	ACCCA	4	75	17	non-palindrome				
+contig_2	a	CCAAAT	4	71	5	non-palindrome				
+contig_2	a	GATC	1	679	74	palindrome	GATC	1	679	74
+contig_2	a	GRNGAAGY	5	85	6	non-palindrome				
+contig_3	a	ACCCA	4	58	8	non-palindrome				
+contig_3	a	CCAAAT	4	80	1	non-palindrome				
+contig_3	a	GATC	1	680	29	palindrome	GATC	1	680	29
+contig_3	a	GRNGAAGY	5	83	2	non-palindrome				

nanomotif/datasets/geobacillus-plasmids.motifs.tsv

@@ -0,0 +1,9 @@
+contig	motif	mod_position	mod_type	n_mod	n_nomod	motif_type	motif_complement	mod_position_complement	n_mod_complement	n_nomod_complement
+contig_2	ACCCA	4	a	75	17	non-palindrome				
+contig_2	CCAAAT	4	a	71	5	non-palindrome				
+contig_2	GATC	1	a	679	74	palindrome	GATC	1	679	74
+contig_2	GRNGAAGY	5	a	85	8	non-palindrome				
+contig_3	ACCCA	4	a	58	8	non-palindrome				
+contig_3	CCAAAT	4	a	80	1	non-palindrome				
+contig_3	GATC	1	a	680	29	palindrome	GATC	1	680	29
+contig_3	GRNGAAGY	5	a	83	2	non-palindrome				

nanomotif/main.py

@@ -221,7 +221,7 @@ def score_motifs(args, pl, pileup = None, assembly = None, motifs = None, min_mo
     log.info("Starting nanomotif motif scorer")
     if pileup is None:
         log.info("Loading pileup")
-        pileup =  nm.load_pileup(args.pileup, min_fraction = args.threshold_methylation_general)
+        pileup =  nm.load_pileup(args.pileup, min_fraction = args.threshold_methylation_general, min_coverage = args.threshold_valid_coverage)
     if assembly is None:
         log.info("Loading assembly")
         assembly = nm.load_assembly(args.assembly)
@@ -299,7 +299,7 @@ def bin_consensus(args, pl, pileup = None, assembly = None, motifs = None, motif
         motifs_scored = pl.read_csv(args.motifs_scored, separator="\t")
     if pileup is None:
         log.info("Loading pileup")
-        pileup =  nm.load_pileup(args.pileup, threads = args.threads, min_fraction = args.threshold_methylation_general)
+        pileup =  nm.load_pileup(args.pileup, min_fraction = args.threshold_methylation_general, min_coverage = args.threshold_valid_coverage)
     if assembly is None:
         log.info("Loading assembly")
         assembly = nm.load_assembly(args.assembly)

tests/test_cli_commands.py

@@ -0,0 +1,54 @@
+import pytest
+import subprocess
+import pandas as pd
+import os
+import glob
+import shutil
+
+
+def test_bin_consensus():
+    """
+    """
+    outdir = "tests/cli/test_bin_consensus"
+
+    cmd = [
+        "nanomotif", "bin_consensus",
+        "-t", "1",
+        "nanomotif/datasets/geobacillus-plasmids.assembly.fasta",
+        "nanomotif/datasets/geobacillus-plasmids.pileup.bed",
+        "nanomotif/datasets/geobacillus-plasmids.motifs.tsv",
+        "nanomotif/datasets/geobacillus-contig-bin.tsv",
+        "nanomotif/datasets/geobacillus-plasmids.motifs-scored.tsv",
+        "--out", outdir
+    ]
+
+    result = subprocess.run(cmd)
+
+    # Check that the CLI tool executed successfully
+    assert result.returncode == 0, "CLI tool did not exit successfully"
+
+    #TODO: remove outputdir
+    #
+
+
+def test_score_motifs():
+    """
+    """
+    outdir = "tests/cli/test_score_motifs"
+
+    cmd = [
+        "nanomotif", "score_motifs",
+        "-t", "1",
+        "nanomotif/datasets/geobacillus-plasmids.assembly.fasta",
+        "nanomotif/datasets/geobacillus-plasmids.pileup.bed",
+        "nanomotif/datasets/geobacillus-plasmids.motifs.tsv",
+        "--out", outdir
+    ]
+
+    result = subprocess.run(cmd)
+
+    # Check that the CLI tool executed successfully
+    assert result.returncode == 0, "CLI tool did not exit successfully"
+
+    #TODO: remove outputdir
+