bug fixed: ignored parameter in stageone

README.md

@@ -10,9 +10,10 @@ The change log of this version (June, 2022) includes:
 + We modified the 16s estimation process by changing minimap2 to bwa + blastn, as minimap2 does not work well for reads that are super short (e.g. below 100 bp, see [https://github.com/lh3/minimap2/issues/363#issuecomment-473387994](https://github.com/lh3/minimap2/issues/363#issuecomment-473387994)).
 + We fixed the version of diamond to 0.9.24 (and python to 3.7.\*), as the latest version of diamond (2.0.15) will gives ~10% more hits of USCMGs and ARGs. The sensitivity of the newer version of diamond is under evaluation. We hope to remove this constrain in future updates.
 + Bug fixed:
-    + Fixed a bug that caused the worst hits (instead of the best) out of five being picked in stagetwo's blastn.
-    + Fixed a bug in stageone caused USCMG being slightly overestimated.
-    + Fixed a bug in stagetwo that caused some multi-component ARGs hits being ignored.
+    + Fixed a bug that caused the worst hits (instead of the best) being picked in stagetwo's blastx when multiple candidates of ARGs can be found.
+    + Fixed a bug that caused some multi-component ARGs hits being uncounted in stagetwo's aggregation process.
+    + Fixed a bug in stageone that caused USCMG being slightly overestimated.
+    + Fixed a bug in stageone that caused parameters -x -y -v being ignored.
 
 ## Installation
 Conda (osx-64/linux-64):

args_oap/argoap_pipeline_stageone_version3

@@ -39,25 +39,26 @@ USE
 
 #Get all the input parameters
 
-getopts('i:o:n:f:m:x:y:v:hzq');
+getopts('i:o:m:n:f:q:z:x:y:v:h');
 if($opt_h  ||  ($opt_i eq "") ){
 	die "$usage\n";
 }
 
 $opt_n ||= 1;
 $opt_f ||="fq";
 $opt_o ||= ".";
-unless(-d $opt_o){
-	`mkdir $opt_o`;
-}
-
 $opt_x ||= 1e-10;
 $opt_y ||= 3;
 $opt_v ||= 45;
+
 my $eval1 =$opt_x;
 my $eval2 =$opt_y;
 my $did = $opt_v; ##identity for diamond searching
 
+unless(-d $opt_o){
+	`mkdir $opt_o`;
+}
+
 my $logF = "$opt_o/Log.txt";
 my $date = localtime;
 open(LOG, ">$logF") || die "Can not create Log.txt\n";
@@ -75,17 +76,12 @@ my $ggnr85blast ||= "$ublastxdir/DB/gg85.fasta";
 my $KO30DMD ||= "$ublastxdir/DB/all_KO30.fasta.dmnd";
 my $genekolist ||="$ublastxdir/DB/all_KO30_name.list";
 
-
 ##Get all the fastq in input dir
 ##Generate output script files for all fastq
 $date =~ s/\s+/-/g;
 
 my $o_sh = "$opt_o/ublastx_bash_$date.sh";
-unless(-d $opt_o){
-	`mkdir $opt_o`;
-}
 open(BASH, ">$o_sh") || die "output dir can not write:\n"; 
-
 my $count = 1;
 
 ##--------------------------------------------------------------------------------------------------------

args_oap/args_oap.py

@@ -23,10 +23,10 @@ def parse_options(argv):
     optional_one.add_argument('-n', help='number of threads, default 1', default=1)
     optional_one.add_argument('-f', help='the format of processed files, default fq', default='fq')
     optional_one.add_argument('-q', help='quality control of fastq sequences defualt not take effect, set to 1, then will do QC with fastp', default=0)
-    optional_one.add_argument('-z', help='whether the fq files were .gz format, if -z, then firstly gzip -d, default(none) ', default=False, action='store_true')
+    optional_one.add_argument('-z', help='whether the fq files were .gz format, if 1, then firstly gzip -d, default 0', default=0)
     optional_one.add_argument('-x', help='evalue for searching 16S in usearch default 1e-10', default=1e-10)
     optional_one.add_argument('-y', help='evalue for searching universal single copy marker gene default 3', default=3)
-    optional_one.add_argument('-v', help='the identity value for diamond to search the USCMGs default  0.45', default=0.45)
+    optional_one.add_argument('-v', help='the identity value for diamond to search the USCMGs default 45', default=45)
 
     parser_stage_one.set_defaults(func=stage_one)
 

args_oap/stage_one.py

@@ -14,11 +14,10 @@ def stage_one(options):
     '-n', str(options.n),
     '-f', options.f,
     '-q', str(options.q),
+    '-z', str(options.z),
     '-x', str(options.x),
     '-y', str(options.y),
     '-v', str(options.v)]
 
-    if options.z:
-        args.extend(['-z'])
-
     subprocess.call(args)
+