Merge pull request #21 from bodegalab/more_verbosity

Set multiple verbosity level
bodegalab · Sep 19, 2024 · 29e7bb9 · 29e7bb9
2 parents 22ef8a8 + 12a8a7c
commit 29e7bb9
Show file tree

Hide file tree

Showing 5 changed files with 33 additions and 30 deletions.
diff --git a/irescue/count.py b/irescue/count.py
@@ -248,7 +248,7 @@ def run_count(maps_file, features_index, tmpdir, dumpEC, verbose,
             writerr(
                 f'[{taskn}] Run count for cell '
                 f'{cellidx} ({cellbarcode.decode()})',
-                send=verbose
+                level=2, send=verbose
             )
             cellcounts, dump = compute_cell_counts(
                 equivalence_classes=cellmaps,
@@ -258,7 +258,7 @@ def run_count(maps_file, features_index, tmpdir, dumpEC, verbose,
             writerr(
                 f'[{taskn}] Write count for cell '
                 f'{cellidx} ({cellbarcode.decode()})',
-                send=verbose
+                level=1, send=verbose
             )
             # round counts to 3rd decimal point and write to matrix file
             # only if count is at least 0.001
@@ -270,7 +270,7 @@ def run_count(maps_file, features_index, tmpdir, dumpEC, verbose,
                 writerr(
                     f'[{taskn}] Write ECdump for cell '
                     f'{cellidx} ({cellbarcode.decode()})',
-                    send=verbose
+                    level=1, send=verbose
                 )
                 # reverse features index to get names back
                 findex = dict(zip(features_index.values(),

diff --git a/irescue/main.py b/irescue/main.py
@@ -72,10 +72,10 @@ def parseArguments():
                         "presence in bam file.")
     parser.add_argument('--keeptmp', action='store_true',
                         help="Keep temporary files under <output_dir>/tmp.")
-    parser.add_argument('-v', '--verbose', action='store_true',
-                        help="Writes a lot of stuff to stderr, such as "
-                        "chromosomes as they are mapped and cell barcodes "
-                        "as they are processed.")
+    parser.add_argument('-v', '--verbose', action='count', default=0,
+                        help="Writes additional logging to stderr. "
+                        "Use once for normal verbosity (-v), "
+                        "twice for debugging (-vv).")
     parser.add_argument('-V', '--version', action='version',
                         version='%(prog)s {}'.format(__version__),
                         help="Print software's version and exit.")
@@ -156,7 +156,7 @@ def main():
     writerr(
         "Computing overlap between reads and TEs coordinates in the "
         "following references: {}".format(', '.join(chrNames)),
-        send=args.verbose
+        level=1, send=args.verbose
     )
     isecFun = partial(
         isec, args.bam, regions, whitelist, args.cb_tag, args.umi_tag,
@@ -214,7 +214,7 @@ def main():
         writerr(f'Writing Equivalence Classes to {ecdump_file}')
 
     if not args.keeptmp:
-        writerr(f'Cleaning up temporary files.', send=args.verbose)
+        writerr(f'Cleaning up temporary files.', level=1, send=args.verbose)
         rmtree(dirs['tmp'])
 
     writerr('Done.')
diff --git a/irescue/map.py b/irescue/map.py
@@ -26,7 +26,8 @@ def checkIndex(bamFile, verbose):
                 writerr('BAM indexing done.')
         else:
             if verbose:
-                writerr(f'Found index for BAM file {bamFile}', send=verbose)
+                writerr(f'Found index for BAM file {bamFile}',
+                        level=1, send=verbose)
 
 
 # Check repeatmasker regions bed file format. Download if not provided.
@@ -208,11 +209,11 @@ def isec(bamFile, bedFile, whitelist, CBtag, UMItag, bpOverlap, fracOverlap,
     cmd += ' print qname[n-1]"\\t"qname[n]"\\t"qname[1]"\\t"$16 }\' '
     cmd += f' | gzip > {isecFile}'
 
-    writerr(f'Extracting {chrom} reference', send=verbose)
+    writerr(f'Extracting {chrom} reference', level=2, send=verbose)
     run_shell_cmd(cmd0)
-    writerr(f'Intersecting alignments with {chrom} reference', send=verbose)
+    writerr(f'Mapping alignments to {chrom}', level=1, send=verbose)
     run_shell_cmd(cmd)
-    writerr(f'Finished mapping {chrom}', send=verbose)
+    writerr(f'Mapped {chrom}', level=1, send=verbose)
 
     return isecFile
 
@@ -247,7 +248,7 @@ def chrcat(filesList, threads, outdir, tmpdir, bedtools, verbose):
     cmd2 += ' print $1"\\t"gensub(/#.+/,"",1,$1)"\\tGene Expression" }\' '
     cmd2 += f' | LC_ALL=C sort -u | gzip > {features_file} '
 
-    writerr('Concatenating mappings', send=verbose)
+    writerr('Concatenating mappings', level=1, send=verbose)
     run_shell_cmd(cmd0)
     if getlen(mappings_file) == 0:
         writerr(

diff --git a/irescue/misc.py b/irescue/misc.py
@@ -58,7 +58,7 @@ def check_requirement(cmd, required_version, parser, verbose):
             else:
                 writerr(
                     f"Found {cmd} version {version}. Proceeding.",
-                    send=verbose
+                    level=1, send=verbose
                 )
         except:
             writerr(
@@ -68,7 +68,7 @@ def check_requirement(cmd, required_version, parser, verbose):
             )
 
 # Small function to write a message to stderr with timestamp
-def writerr(msg, error=False, send=True):
+def writerr(msg, error=False, level=0, send=0):
     """
     Write a message to stderr with timestamp.
 
@@ -78,10 +78,12 @@ def writerr(msg, error=False, send=True):
          Message to write to stderr.
     error: bool
            Set True if the message is an error (write with sys.exit).
-    send: bool
-          Decides if the message should be sent (useful for verbose messages).
+    verbosity: int
+            Verbosity level of the message.
+    send: int
+          If >=verbosity, the message will be sent.
     """
-    if send:
+    if send>=level or error:
         timelog = datetime.now().strftime("%Y/%m/%d - %H:%M:%S")
         message = f'[{timelog}] '
         if not msg[-1]=='\n':
@@ -158,7 +160,7 @@ def check_tags(
         writerr(
             f"Testing bam file for {CBtag} and {UMItag} tags presence. "
             "Will stop at the first occurrence.",
-            send=verbose
+            level=1, send=verbose
         )
         for read in f:
             if nLines and c >= nLines:

diff --git a/tests/test.yml b/tests/test.yml
@@ -1,5 +1,5 @@
 - name: base
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -v
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -vv
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 1a74fa12e65ac1703bbe61282854f151
@@ -11,7 +11,7 @@
 - name: genome
   tags:
     - genome
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -g test --keeptmp -v
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -g test --keeptmp -vv
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 1a74fa12e65ac1703bbe61282854f151
@@ -23,7 +23,7 @@
 - name: multi
   tags:
     - multi
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz -p 2 --keeptmp -v
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz -p 2 --keeptmp -vv
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 1a74fa12e65ac1703bbe61282854f151
@@ -35,7 +35,7 @@
 - name: whitelist
   tags:
     - whitelist
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz -w ./tests/data/whitelist.txt --keeptmp -v
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz -w ./tests/data/whitelist.txt --keeptmp -vv
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 95dccc15cbee4feeeae2fbce4d7b41ad
@@ -48,7 +48,7 @@
   tags:
     - multi
     - whitelist
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz -w ./tests/data/whitelist.txt --keeptmp -v -p 2
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz -w ./tests/data/whitelist.txt --keeptmp -vv -p 2
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 95dccc15cbee4feeeae2fbce4d7b41ad
@@ -60,7 +60,7 @@
 - name: ecdump
   tags:
     - ecdump
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -v --dump-ec
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -vv --dump-ec
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 1a74fa12e65ac1703bbe61282854f151
@@ -75,7 +75,7 @@
   tags:
     - multi
     - ecdump
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -v -p 2 --dump-ec
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -vv -p 2 --dump-ec
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 1a74fa12e65ac1703bbe61282854f151
@@ -89,7 +89,7 @@
 - name: bp
   tags:
     - bp
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -v --min-bp-overlap 10
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -vv --min-bp-overlap 10
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 7433e88e94aec2f16a20459275188f1f
@@ -101,7 +101,7 @@
 - name: fraction
   tags:
     - fraction
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -v --min-fraction-overlap 0.5
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -vv --min-fraction-overlap 0.5
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 4de44d3e4a851392a48ccabfee5bb6fc
@@ -114,7 +114,7 @@
   tags:
     - bp
     - fraction
-  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -v --min-bp-overlap 10 --min-fraction-overlap 0.5
+  command: irescue -b ./tests/data/Aligned.sortedByCoord.out.bam -r ./tests/data/rmsk.bed.gz --keeptmp -vv --min-bp-overlap 10 --min-fraction-overlap 0.5
   files:
     - path: "irescue_out/counts/barcodes.tsv.gz"
       md5sum: 4de44d3e4a851392a48ccabfee5bb6fc