Exchanged prinseq for HTStream for QC

config/default.yml

@@ -59,12 +59,21 @@ salmon:
   quant_flags: '--gcBias'
 
 #
-# The PRINSEQ section
-#
-# flags - flags to add to the PRINSEQ call while cleaning fastq files
-#
-prinseq:
-  flags: '-derep 1 -custom_params "A 8"'
+# The HTStream section
+#
+# chain - actions of HTStream (will be prefixed with hts_) and their flags
+#
+htstream:
+  actions:
+    - PolyATTrim
+    - QWindowTrim
+    - SuperDeduper
+    - Stats
+  flags:
+    - '-StOA --min-trim 8 --max-mismatch 3'
+    - '-StOA --window-size 20 --avg-qual 20 --min-length 50'
+    - '-StOA --start 1 --length 50'
+    - '-SFfgA --notes stats_after_qc'
 
 #
 # The fastq-multx section

environment.yml

@@ -7,5 +7,5 @@ channels:
 dependencies:
   - snakemake
   - salmon
-  - prinseq
+  #- htstream
   - fastq-multx

src/clean.snake

@@ -1,16 +1,21 @@
+def construct_htstream_partials(wildcards, log):
+  template = 'hts_{prog} -L {log}/htstream/{sample}.txt {flags}'
+  partials = [template.format(prog = k, log = log, flags = v, sample = wildcards.sample) for k, v in list(zip(config['htstream']['actions'], config['htstream']['flags']))]
+  return(' | '.join(partials))
 
-rule prinseq_clean_se_gz:
+rule htstream_clean_se_gz:
   version: "1.0"
   input:
     '{out}/{sample}.fastq.gz'
   output:
-    clean = '{out}/{sample}.clean.fastq.gz',
-    grubby = '{out}/{sample}.grubby.fastq.gz',
+    '{out}/{sample}.clean.fastq.gz'
   params:
-    flags = config['prinseq']['flags'] if 'flags' in config.get('prinseq', {}) else ''
-  message: 'Cleaning {input} using prinseq-lite.'
-  log: join(config['dirs']['log'], 'prinseq', 'clean-se.log')
+    call = lambda wildcards: construct_htstream_partials(wildcards, config['dirs']['log'])
+  message: 'Cleaning {input} using HTStream.'
+  log: join(config['dirs']['log'], 'htstream', '{sample}.txt')
+  threads: 4
   shell:
     """
-    zcat {input} | prinseq-lite.pl -fastq -out_good {wildcards.out}/{wildcards.sample}.clean -out_bad {wildcards.out}/{wildcards.sample}.grubby -graph_data {log} {params.flags}
+    hts_Stats -U {input} -L {log} -tO | {params.call} -p {wildcards.out}/{wildcards.sample}
+    mv {wildcards.out}/{wildcards.sample}_SE.fastq.gz {output}
     """