parsing maxambiguity and minlength options as flagged in issue #405

pangolin/command.py

@@ -162,7 +162,7 @@ def main(sysargs = sys.argv[1:]):
     config[KEY_OUTFILE] = io.set_up_outfile(args.outfile, config[KEY_OUTFILE],config[KEY_OUTDIR])
     io.set_up_tempdir(args.tempdir,args.no_temp,cwd,config[KEY_OUTDIR], config)
     config[KEY_ALIGNMENT_FILE],config[KEY_ALIGNMENT_OUT] = io.parse_alignment_options(args.alignment, config[KEY_OUTDIR], config[KEY_TEMPDIR],args.alignment_file, config[KEY_ALIGNMENT_FILE])
-
+    parse_qc_thresholds(args.maxambig, args.minlen, config[KEY_REFERENCE_FASTA], config)
     config[KEY_QUERY_FASTA] = io.find_query_file(cwd, config[KEY_TEMPDIR], args.query)
 
     io.quick_check_query_file(cwd, args.query, config[KEY_QUERY_FASTA])

pangolin/utils/config.py

@@ -15,7 +15,6 @@
 
 KEY_DATADIR="datadir"
 
-KEY_MINLEN="minlen"
 KEY_MAXAMBIG="maxambig"
 KEY_TRIM_START="trim_start"
 KEY_TRIM_END="trim_end"

pangolin/utils/initialising.py

@@ -2,6 +2,7 @@
 import sys
 import itertools
 from distutils.version import LooseVersion
+from Bio import SeqIO
 
 import pangolin.utils.custom_logger as custom_logger
 from pangolin.utils.log_colours import green,cyan
@@ -33,7 +34,6 @@ def setup_config_dict(cwd):
 
             KEY_DATADIR:None,
 
-            KEY_MINLEN: 25000,
             KEY_MAXAMBIG: 0.3,
             KEY_TRIM_START:265, # where to pad to using datafunk
             KEY_TRIM_END:29674, # where to pad after using datafunk
@@ -176,6 +176,35 @@ def setup_data(datadir_arg,analysis_mode, config):
     config[KEY_DATADIR] = datadir
     config[KEY_CONSTELLATION_FILES] = constellation_files
 
+def parse_qc_thresholds(maxambig, minlen, reference_fasta, config):
+
+    if maxambig:
+        maxambig = float(maxambig)
+        if maxambig <=1 and maxambig >= 0:
+            config[KEY_MAXAMBIG] = maxambig
+        else:
+            sys.stderr.write(cyan(f'Error: `--max-ambiguity` should be a float between 0 and 1.\n'))
+            sys.exit(-1)
+
+    if minlen:
+        minlen = float(minlen)
+        reflen = 0
+        for record in SeqIO.parse(reference_fasta,"fasta"):
+            reflen = len(record)
+
+        if minlen>reflen:
+            sys.stderr.write(cyan(f'Error: `--min-length` should be less than the length of the reference: {ref_len}.\n'))
+            sys.exit(-1)
+        else:
+            new_maxambig = round(1-(minlen/reflen), 3)
+            print(f"Converting minimum length of {minlen} to maximum ambiguity of {new_maxambig}.")
+            if new_maxambig > config[KEY_MAXAMBIG]:
+                config[KEY_MAXAMBIG] = new_maxambig
+
+    print(green(f"Maximum ambiguity allowed is {config[KEY_MAXAMBIG]}.\n****"))
+
+
+
 
 def print_alias_file_exit(alias_file):
     with open(alias_file, 'r') as handle:

pangolin/utils/io_parsing.py

@@ -11,6 +11,7 @@
 
 from pangolin.utils.config import *
 
+
 def find_query_file(cwd, tempdir, query_arg):
     if len(query_arg) > 1:
         print(cyan(f"Error: Too many query (input) fasta files supplied: {query_arg}\nPlease supply one only."))