changeset 1:c43f4b80f5a9 draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/nanocompore commit 652a872061324ba1073bfa286777ffeefa352671"

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/NanocomporeDB_process.py	Mon Jun 08 14:36:18 2020 -0400
@@ -0,0 +1,60 @@
+#!/usr/bin/env python
+
+import argparse
+import os
+
+from nanocompore.SampCompDB import SampCompDB
+
+
+def is_valid_file(file_name):
+    if os.path.isfile(file_name):
+        return os.path.abspath(file_name)
+    else:
+        raise FileNotFoundError(os.path.abspath(file_name))
+
+
+def is_valid_directory(dir_name):
+    if os.path.isdir(dir_name):
+        return os.path.abspath(dir_name)
+    else:
+        raise NotADirectoryError(os.path.abspath(dir_name))
+
+
+if __name__ == '__main__':
+
+    parser = argparse.ArgumentParser(
+        description='save nanocompre sampcomp \
+            results as interval outputs \
+            \nSample call: \"python Nannocompore-plot.py --db-path \
+            ./out_SampComp.db --ref-fasta ref.fa --annotation-bed annot.bed \
+            --out-dir ./plots/')
+
+    parser.add_argument('--ref-fasta', required=True, type=is_valid_file,
+                        help='The reference genome  used for read alignment.')
+    parser.add_argument('--db-path', default="./out_SampComp.db", type=str,
+                        help='Path to the SampCompDB database path prefix.')
+    parser.add_argument('--annotation-bed', required=False, type=is_valid_file,
+                        help='BED file containing the annotation of the transcriptome used as reference when mapping')
+    parser.add_argument('--pvalue-types', type=str,
+                        default='GMM_logit_pvalue,KS_dwell_pvalue,KS_intensity_pvalue',
+                        help='path to the annotations')
+    parser.add_argument('--bedgraph', default=False,
+                        help='write output in BEDGRAPH format instead of BED')
+    parser.add_argument('--pvalue-threshold', default=1.0,
+                        help='Maximum reported p-value.')
+    parser.add_argument('--out-dir', default="./", type=is_valid_directory,
+                        help='path the plotting output directory.')
+
+    args = parser.parse_args()
+
+    db = SampCompDB(args.db_path, fasta_fn=args.ref_fasta,
+                    bed_fn=args.annotation_bed)
+    print(db)
+    print("DB read ids:", db.ref_id_list)
+
+    if args.annotation_bed:
+        for pt in args.pvalue_types.split(','):
+            print("bedgraph output for p-value type:", pt)
+            db.save_to_bed(output_fn='{}/{}.bedgraph'.format(args.out_dir, pt),
+                           pvalue_field=pt, pvalue_thr=args.pvalue_threshold,
+                           bedgraph=args.bedgraph)

--- a/macros.xml	Tue May 05 06:57:10 2020 -0400
+++ b/macros.xml	Mon Jun 08 14:36:18 2020 -0400
@@ -6,6 +6,7 @@
     <xml name="requirements">
         <requirements>
             <requirement type="package" version="@TOOL_VERSION@">nanocompore</requirement>
+            <requirement type="package" version="1.32">tar</requirement>
         </requirements>
     </xml>
     <xml name="citations">

--- a/sampcomp.xml	Tue May 05 06:57:10 2020 -0400
+++ b/sampcomp.xml	Mon Jun 08 14:36:18 2020 -0400
@@ -60,7 +60,7 @@
             <param name="file" type="data" format="tabular" label="Select NanopolishComp file" help="(--file_list2)"/>
             <param name="index" type="data" format="tabular" label="Select index file"/>
         </repeat>
-        <param argument="--fasta" type="data" format="fasta" label="Select mapping file"/>
+        <param argument="--fasta" type="data" format="fasta" label="Select reference mapping fasta"/>
 
         <section name="ap" title="Advanced parameters">
             <param argument="--bed" type="data" format="bed" optional="true" label="Select mapping file with annotation of transcriptome"/>
@@ -131,7 +131,7 @@
             </output>
             <output name="out_db">
                 <assert_contents>
-                    <has_size value="5408256"/>
+                    <has_size value="5416960" delta="10000"/>
                 </assert_contents>
             </output>
         </test>
@@ -176,7 +176,7 @@
             </output>
             <output name="out_db">
                 <assert_contents>
-                    <has_size value="5410304"/>
+                    <has_size value="5416960" delta="10000"/>
                 </assert_contents>
             </output>
             <output name="out_log">

Binary file test-data/DB.tar has changed

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/GMM_logit_pvalue.bedgraph	Mon Jun 08 14:36:18 2020 -0400
@@ -0,0 +1,2 @@
+chr	22102	22107	chr_AGGAA	0	+
+chr	22103	22108	chr_GGAAA	1.0472150418546853	+

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/KS_dwell_pvalue.bedgraph	Mon Jun 08 14:36:18 2020 -0400
@@ -0,0 +1,2 @@
+chr	22102	22107	chr_AGGAA	0.037294307603991536	+
+chr	22103	22108	chr_GGAAA	0.037294307603991536	+

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/KS_intensity_pvalue.bedgraph	Mon Jun 08 14:36:18 2020 -0400
@@ -0,0 +1,2 @@
+chr	22102	22107	chr_AGGAA	0.5940708247528699	+
+chr	22103	22108	chr_GGAAA	0.5940708247528699	+

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/annot.bed	Mon Jun 08 14:36:18 2020 -0400
@@ -0,0 +1,1 @@
+chr	1	30000	chr	0	+