add SAMPLE > CHEMISTRY > DEFAULT config logic and simplify STAR argum…

…ents

Snakefile

@@ -24,7 +24,7 @@ def _get_config(sample, item):
     return SAMPLES[sample][item]
   except KeyError:
     try:
-      return CHEMISTRY[item]
+      return CHEMISTRY[SAMPLES[sample]["chemistry"]][item]
     except KeyError:
       return DEFAULTS[item]
 

chemistry.yaml

@@ -1,47 +1,38 @@
 chromiumV3_illumina:
   filter_R1_length: 58
-  bc_cut: ""
-  STAR_R1: "[WHITELIST_V3,\"--soloUMIlen 12 --clip5pNbases 58 0 --soloCBstart 1 --soloCBlen 16 --soloUMIstart 17 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2\"]"
-  STAR_R2: "[WHITELIST_V3,\"--soloUMIlen 12\"]"
-  STAR_paired: "--alignEndsProtrude 58 ConcordantPair"
+  bc_whitelist: WHITELIST_V3
+  STAR_R1: --soloUMIlen 12 --clip5pNbases 58 0 --soloCBstart 1 --soloCBlen 16 --soloUMIstart 17 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2
+  STAR_R2: --soloUMIlen 12
 chromiumV3_element:
   filter_R1_length: False
-  bc_cut: ""
-  STAR_R1: "[WHITELIST_V3,\"--soloUMIlen 12 --clip5pNbases 48 0 --soloCBstart 1 --soloCBlen 16 --soloUMIstart 17\"]"
-  STAR_R2: "[WHITELIST_V3,\"--soloUMIlen 12\"]"
-  STAR_paired: "--alignEndsProtrude 58 ConcordantPair"
+  bc_whitelist: WHITELIST_V3
+  STAR_R1: --soloUMIlen 12 --clip5pNbases 48 0 --soloCBstart 1 --soloCBlen 16 --soloUMIstart 17
+  STAR_R2: --soloUMIlen 12
 chromiumV3_ultima:
   filter_R1_length: 58
-  bc_cut: ""
-  STAR_R1: "[WHITELIST_V3,\"--soloUMIlen 9 --clip5pNbases 58 --soloCBstart 23 --soloCBlen 16 --soloUMIstart 39\"]"
+  bc_whitelist: WHITELIST_V3
+  STAR_R1: --soloUMIlen 9 --clip5pNbases 58 --soloCBstart 23 --soloCBlen 16 --soloUMIstart 39
 chromiumV2_illumina:
   filter_R1_length: 56
-  bc_cut: ""
-  STAR_R1: "[WHITELIST_V2,\"--soloUMIlen 10 --clip5pNbases 56 0 --soloCBstart 1 --soloCBlen 16 --soloUMIstart 17 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2\"]"
-  STAR_R2: "[WHITELIST_V2,\"--soloUMIlen 10\"]"
-  STAR_paired: "--alignEndsProtrude 56 ConcordantPair"
+  bc_whitelist: WHITELIST_V2
+  STAR_R1: --soloUMIlen 10 --clip5pNbases 56 0 --soloCBstart 1 --soloCBlen 16 --soloUMIstart 17 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2
+  STAR_R2: --soloUMIlen 10
 dropseq_illumina:
   filter_R1_length: 50
-  bc_cut: ""
-  STAR_R1: "None --soloUMIlen 8 --clip5pNbases 50 0 --soloCBstart 1 --soloCBlen 12 --soloUMIstart 13 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2"
-  STAR_R2: "None --soloUMIlen 8 --soloCBstart 1 --soloCBlen 12 --soloUMIstart 13"
-  STAR_paired: "--alignEndsProtrude 50 ConcordantPair"
+  STAR_R1: --soloUMIlen 8 --clip5pNbases 50 0 --soloCBstart 1 --soloCBlen 12 --soloUMIstart 13 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2
+  STAR_R2: --soloUMIlen 8 --soloCBstart 1 --soloCBlen 12 --soloUMIstart 13
 microwellseq_illumina:
   filter_R1_length: 54
-  bc_cut: "CGACTCACTACAGGG...TCGGTGACACGATCG"
-  STAR_R1: "None --soloUMIlen 6 --clip5pNbases 54 0 --soloCBstart 1 --soloCBlen 18 --soloUMIstart 19 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2"
-  STAR_R2: "None --soloUMIlen 6 --soloCBstart 1 --soloCBlen 18 --soloUMIstart 19"
-  STAR_paired: "--alignEndsProtrude 54 ConcordantPair"
+  bc_cut: CGACTCACTACAGGG...TCGGTGACACGATCG
+  STAR_R1: --soloUMIlen 6 --clip5pNbases 54 0 --soloCBstart 1 --soloCBlen 18 --soloUMIstart 19 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2
+  STAR_R2:  --soloUMIlen 6 --soloCBstart 1 --soloCBlen 18 --soloUMIstart 19
 bd_illumina:
   filter_R1_length: 53
-  bc_cut: "ACTGGCCTGCGA...GGTAGCGGTGACA"
-  STAR_R1: "None --soloUMIlen 8 --clip5pNbases 53 0 --soloCBstart 1 --soloCBlen 27 --soloUMIstart 28 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2"
-  STAR_R2: "None --soloUMIlen 8 --soloCBstart 1 --soloCBlen 27 --soloUMIstart 28"
-  STAR_paired: "--alignEndsProtrude 53 ConcordantPair"
+  bc_cut: ACTGGCCTGCGA...GGTAGCGGTGACA
+  STAR_R1: --soloUMIlen 8 --clip5pNbases 53 0 --soloCBstart 1 --soloCBlen 27 --soloUMIstart 28 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2
+  STAR_R2: --soloUMIlen 8 --soloCBstart 1 --soloCBlen 27 --soloUMIstart 28
 indrop_illumina:
   filter_R1_length: 32
-  bc_cut: ""
-  STAR_R1: "None --soloUMIlen 6 --clip5pNbases 32 0 --soloCBstart 1 --soloCBlen 8 --soloUMIstart 9 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2"
-  STAR_R2: "None --soloUMIlen 6 --soloCBstart 1 --soloCBlen 8 --soloUMIstart 9"
-  STAR_paired: "--alignEndsProtrude 32 ConcordantPair"
+  STAR_R1: --soloUMIlen 6 --clip5pNbases 32 0 --soloCBstart 1 --soloCBlen 8 --soloUMIstart 9 --outFilterMultimapNmax 1  --outFilterMismatchNmax 999  --outFilterMismatchNoverReadLmax 0.2
+  STAR_R2: --soloUMIlen 6 --soloCBstart 1 --soloCBlen 8 --soloUMIstart 9
 

config.yaml

@@ -34,23 +34,23 @@ POLYA_SITES:
 #   "SAF"
 
 DEFAULTS:
-# default config options, overridden by SAMPLES
+# default config options, overridden by SAMPLES and CHEMISTRY definitions
   chemistry: chromiumV3_illumina
   alignments:
     - R2
     - paired
-  extra_args: ""
+  STAR_paired: --alignEndsProtrude 58 ConcordantPair
 
 SAMPLES:
 # per-sample labels, FASTQ basenames, and config options
 # required:
 #   basename
 # optional:
 #   chemisty
-#   platform
 #   alignments
 #   cutadapt_args
-#   star_args
+#   star_args (STAR_R1, STAR_R2, STAR_paired)
+#   extra star_args (STAR_R1_extra_args, STAR_R2_extra_args, STAR_paired_extra_args)
   test:
     basename: sample
     chemistry: chromiumV2_illumina

rules/cutadapt_star.snake

@@ -15,33 +15,34 @@ def _get_fq_paths(wildcards):
 
 """ Process complex barcodes into simple, if needed"""
 def _get_bc_cut(wildcards):
-    args = CHEMISTRY[_get_config(wildcards.sample, "chemistry")]["bc_cut"]
-    return args
+    try:
+      return CHEMISTRY[_get_config(wildcards.sample, "chemistry")]["bc_cut"]
+    except KeyError:
+      return ""
+
+""" Extract per-capture chemistry from gex libs (base function)"""
+def _get_chem_version(sample, alignment):
+    try:
+      whitelist = eval(_get_config(sample, "bc_whitelist"))
+    except KeyError:
+      whitelist = "None"
+    try:
+      extra_args = _get_config(sample, alignment+"_extra_args")
+    except KeyError:
+      extra_args = ""
+    return [whitelist, _get_config(sample, alignment), extra_args]
 
 """ Extract per-capture chemistry from gex libs (R2)"""
 def _get_chem_version_R2(wildcards):
-    try:
-      return eval(CHEMISTRY[_get_config(wildcards.sample, "chemistry")]["STAR_R2"])
-    except SyntaxError:
-      return CHEMISTRY[_get_config(wildcards.sample, "chemistry")]["STAR_R2"]
+    return _get_chem_version(wildcards.sample, "STAR_R2")
 
 """ Extract per-capture chemistry from gex libs (R1/paired)"""
 def _get_chem_version_R1(wildcards):
-    try:
-      return eval(CHEMISTRY[_get_config(wildcards.sample, "chemistry")]["STAR_R1"])
-    except SyntaxError:
-      return CHEMISTRY[_get_config(wildcards.sample, "chemistry")]["STAR_R1"]
+    return _get_chem_version(wildcards.sample, "STAR_R1")
 
 """ Extract per-capture extra arguments for gex paired alignments """
 def _get_chem_version_paired(wildcards):
-    try:
-      return eval(CHEMISTRY[_get_config(wildcards.sample, "chemistry")]["STAR_paired"])
-    except SyntaxError:
-      return CHEMISTRY[_get_config(wildcards.sample, "chemistry")]["STAR_paired"]
-
-""" Extract per-capture extra arguments for gex libs """
-def _get_extra_args(wildcards):
-    return _get_config(wildcards.sample, "extra_args")
+    return _get_chem_version(wildcards.sample, "STAR_R1") + [_get_config(wildcards.sample, "STAR_paired")]
 
 """ This rule trims R1-only libraries """
 rule cutadapt_R1:
@@ -134,7 +135,6 @@ rule starsolo_R1:
       "{results}/{sample}/{sample}_R1_Aligned.sortedByCoord.out.bam"
     params:
       chemistry = _get_chem_version_R1,
-      extra_args = _get_extra_args,
       out_dir = "{results}/{sample}/{sample}_R1_",
       job_name = "star_R1"
     log: "{results}/logs/{sample}_star_R1.out"
@@ -157,7 +157,7 @@ rule starsolo_R1:
       --soloFeatures Gene \
       --alignMatesGapMax 100000 \
       --alignIntronMax 100000 \
-      --soloCellFilter None {params.extra_args} \
+      --soloCellFilter None \
       --outFileNamePrefix {params.out_dir} \
       --outSAMtype BAM SortedByCoordinate \
       --limitBAMsortRAM 48000000000 \
@@ -175,7 +175,6 @@ rule starsolo_R2:
       "{results}/{sample}/{sample}_R2_Aligned.sortedByCoord.out.bam"
     params:
       chemistry = _get_chem_version_R2,
-      extra_args = _get_extra_args,
       out_dir = "{results}/{sample}/{sample}_R2_", 
       job_name = "star_R2"
     log: "{results}/logs/{sample}_star_R2.out"
@@ -197,7 +196,7 @@ rule starsolo_R2:
       --clipAdapterType CellRanger4 \
       --genomeDir {STAR_INDEX} \
       --soloFeatures Gene \
-      --soloCellFilter None {params.extra_args} \
+      --soloCellFilter None \
       --outFileNamePrefix {params.out_dir} \
       --outSAMtype BAM SortedByCoordinate \
       --limitBAMsortRAM 48000000000 \
@@ -215,9 +214,7 @@ rule starsolo_paired:
     output:
       "{results}/{sample}/{sample}_paired_Aligned.sortedByCoord.out.bam"
     params:
-      chemistry = _get_chem_version_R1,
-      star_args_paired = _get_chem_version_paired,
-      extra_args = _get_extra_args,
+      chemistry = _get_chem_version_paired,
       out_dir = "{results}/{sample}/{sample}_paired_", 
       job_name = "star_paired"
     log: "{results}/logs/{sample}_star_paired.out"
@@ -234,14 +231,13 @@ rule starsolo_paired:
       --readFilesCommand gunzip -c \
       --runThreadN 12 \
       --soloCBwhitelist {params.chemistry} \
-      {params.star_args_paired} \
       --soloBarcodeMate 1 \
       --soloStrand Reverse \
       --genomeDir {STAR_INDEX} \
       --soloFeatures Gene \
       --alignMatesGapMax 100000 \
       --alignIntronMax 100000 \
-      --soloCellFilter None {params.extra_args} \
+      --soloCellFilter None \
       --outFileNamePrefix {params.out_dir} \
       --outSAMtype BAM SortedByCoordinate \
       --limitBAMsortRAM 48000000000 \