Add fastp and update docs

aviary/aviary.py

@@ -328,7 +328,7 @@ def main():
 
     qc_group.add_argument(
         '--min-mean-q', '--min_mean_q',
-        help='Minimum mean quality threshold',
+        help='Minimum long read mean quality threshold',
         dest="min_mean_q",
         default=10
     )
@@ -340,6 +340,51 @@ def main():
         default=100
     )
 
+    qc_group.add_argument(
+        '--min-short-read-length', '--min_short_read_length',
+        help='Minimum length of short reads to be kept',
+        dest="min_short_read_length",
+        default=15
+    )
+
+    qc_group.add_argument(
+        '--max-short-read-length', '--max_short_read_length',
+        help='Maximum length of short reads to be kept, 0 = no maximum',
+        dest="max_short_read_length",
+        default=0
+    )
+
+    qc_group.add_argument(
+        '--disable-adpater-trimming', '--disable_adpater_trimming',
+        help='Disable adapter trimming of short reads',
+        type=str2bool,
+        nargs='?',
+        const=True,
+        dest="disable_adapter_trimming",
+        default=False
+    )
+
+    qc_group.add_argument(
+        '--unqualified-percent-limit', '--unqualified_percent_limit',
+        help='how many percents of bases are allowed to be unqualified. Default 40 means 40 percent',
+        dest="unqualified_percent_limit",
+        default=40
+    )
+
+    qc_group.add_argument(
+        '--quality-cutoff', '--quality_cutoff',
+        help='The short read quality value that a base is qualified. Default 15 means phred quality >=Q15 is qualified.',
+        dest="quality_cutoff",
+        default=15
+    )
+
+    qc_group.add_argument(
+        '--extra-fastp-params', '--extra_fastp_params',
+        help='Extra parameters to pass to fastp, supply as a single string e.g. --extra-fastp-params "-V -e 10"',
+        dest="extra_fastp_params",
+        default=''
+    )
+
     qc_group.add_argument(
         '--skip-qc', '--skip_qc',
         help='Skip quality control steps',

aviary/envs/minimap2.yaml

@@ -5,4 +5,5 @@ channels:
 dependencies:
   - samtools = 1.9
   - minimap2 = 2.18
+  - fastp
   - pigz

aviary/modules/processor.py

@@ -148,13 +148,26 @@ def __init__(self,
             self.min_mean_q = args.min_mean_q
             self.keep_percent = args.keep_percent
             self.skip_qc = args.skip_qc
+            self.min_short_read_size = args.min_short_read_length
+            self.max_short_read_size = args.max_short_read_length
+            self.disable_adapter_trimming = args.disable_adapter_trimming
+            self.unqualified_percent_limit = args.unqualified_percent_limit
+            self.quality_cutoff = args.quality_cutoff
+            self.extra_fastp_params = args.extra_fastp_params
         except AttributeError:
             self.reference_filter = 'none'
             self.gold_standard = 'none'
             self.min_read_size = 0
             self.min_mean_q = 0
             self.keep_percent = 100
             self.skip_qc = False
+            self.min_short_read_size = 0
+            self.max_short_read_size = 0
+            self.disable_adapter_trimming = False
+            self.unqualified_percent_limit = 0
+            self.quality_cutoff = 0
+            self.extra_fastp_params = 'none'
+
 
         try:
             self.gsa_mappings = args.gsa_mappings
@@ -321,6 +334,12 @@ def make_config(self):
         conf["min_read_size"] = self.min_read_size
         conf["min_mean_q"] = self.min_mean_q
         conf["keep_percent"] = self.keep_percent
+        conf["min_short_read_size"] = self.min_short_read_size
+        conf["max_short_read_size"] = self.max_short_read_size
+        conf["disable_adapter_trimming"] = self.disable_adapter_trimming
+        conf["unqualified_percent_limit"] = self.unqualified_percent_limit
+        conf["quality_cutoff"] = self.quality_cutoff
+        conf["extra_fastp_params"] = self.extra_fastp_params
         conf["skip_qc"] = self.skip_qc
         conf["gsa"] = self.gold_standard
         conf["gsa_mappings"] = self.gsa_mappings

aviary/modules/quality_control/qc.smk

@@ -16,6 +16,12 @@ rule qc_short_reads:
         fastq = "data/short_reads.fastq.gz",
         filtered = "data/short_filter.done"
     params:
+        disable_adapter_trimming = config["disable_adapter_trimming"],
+        min_length = config['min_short_read_size'],
+        max_length = config['max_short_read_size'],
+        quality_cutoff = config['quality_cutoff'],
+        unqualified_percent_limit = config['unqualified_percent_limit'],
+        extra_fastp_params = config['extra_fastp_params'],
         coassemble = config["coassemble"],
         reference_filter = [] if "none" in config["reference_filter"] else config["reference_filter"],
         skip_qc = config["skip_qc"]