Fix handling of pre-corrected reads, always operate in HPC space

marbl · Mar 12, 2024 · fc890f7 · fc890f7
1 parent fe736ea
commit fc890f7
Show file tree

Hide file tree

Showing 3 changed files with 17 additions and 22 deletions.
diff --git a/src/Snakefiles/1-buildGraph.sm b/src/Snakefiles/1-buildGraph.sm
@@ -35,7 +35,7 @@ def toFloat(val):
 
 rule buildGraph:
     input:
-        reads     = {rules.correctHiFi.output.fasta} if config['correction_enabled'] == "True" else HIFI_READS,
+        reads     = {rules.correctHiFi.output.fasta},
     output:
         graph     = '1-buildGraph/hifi-resolved.gfa',
         hificov   = '1-buildGraph/hifi_nodecov.csv',
@@ -44,7 +44,6 @@ rule buildGraph:
         err       = '1-buildGraph/buildGraph.err'
     params:
         MBG       = config['MBG'],
-        corenable = config['correction_enabled'],
 
         baseK             = config['mbg_baseK'],
         maxK              = config['mbg_maxK'],
@@ -77,15 +76,8 @@ iopt=""
 
 echo "Building graph with reads:"
 
-if [ "{params.corenable}" = "True" ] ; then
-  echo "  ../{input.reads}"
-  iopt="-i ../{input.reads}"
-else
-  for rf in {HIFI_READS} ; do
-    echo "  \$rf"
-    iopt="\$iopt -i \$rf"
-  done
-fi
+echo "  ../{input.reads}"
+iopt="-i ../{input.reads}"
 
 echo ""
 echo ""

diff --git a/src/Snakefiles/c4-findErrors.sm b/src/Snakefiles/c4-findErrors.sm
@@ -153,16 +153,17 @@ def fixErrorsP(wildcards):
 rule fixErrors:
     input:
         seqstore  = {rules.buildStore.output.seqstore},
-        ovlstore  = {rules.combineOverlaps.output.ovlstore},
-        reds      = fixErrorsI,
-        finished  = {rules.configureFindErrors.output.finished}
+        ovlstore  = {rules.combineOverlaps.output.ovlstore} if config['correction_enabled'] == "True" else {rules.emptyfile.output},
+        reds      = fixErrorsI if config['correction_enabled'] == "True" else {rules.emptyfile.output},
+        finished  = {rules.configureFindErrors.output.finished} if config['correction_enabled'] == "True" else {rules.emptyfile.output}
     output:
         red       = '0-correction/red.red',
         fasta     = '0-correction/hifi-corrected.fasta.gz'
     log:
         err       = "0-correction/fixErrors.err"
     params:
-        paths     = fixErrorsP
+        paths     = fixErrorsP,
+        corenable = config['correction_enabled']
     threads:
         1
     resources:
@@ -181,8 +182,15 @@ echo ""
 echo Combine corrections.
 echo ""
 
+if [ "{params.corenable}" = "False" ] ; then
+   prefix=\\$(echo ../{output.fasta} | sed s/.fasta//)
+   {VERKKO}/bin/sqStoreDumpFASTQ -S ../{input.seqstore} -corrected -compressed -fasta -o \$prefix -nolibname
+   touch ../{output.red}
+   exit 0
+fi
+
 #  red files MUST be in sorted order, which fixErrorsP doesn't return.
-cat find-errors-jobs/*red > red.red
+cat find-errors-jobs/*red > ../{output.red}
 
 #  Use those corrections to correct the reads.
 {VERKKO}/bin/fixErrors \\\\

diff --git a/src/Snakefiles/functions.sm b/src/Snakefiles/functions.sm
@@ -105,12 +105,7 @@ def getMBGMemoryRequest(attempt):              #  Sum the input size, divide by
    if mem < 2.0:
       mem = 0
 
-      if config['correction_enabled'] == "True" and os.path.exists(rules.correctHiFi.output.fasta):
-         mem += getFileBP(rules.correctHiFi.output.fasta)
-      else:
-         for f in HIFI_READS:
-            mem += getFileBP(f)
-
+      mem += getFileBP(rules.correctHiFi.output.fasta)
       mem = mem * 1.10    #  Expand by 10%
       mem = mem / 2       #  MBG is expected to use 1 byte for each 2 bytes of input
       mem = mem / gib     #  Convert to GB