ENH update 03_Quality_control

Author: cocodyq

General_Scripts/03_Quality_control/Readme.md

@@ -50,4 +50,4 @@
 | :---: | :---: | :---: | :---: |
 | 01_merge.py | Merge all the quality control results | GMSC.cluster.tsv.gz rnacode_true_100AA.tsv.xz rnacode_false_100AA.tsv.xz antifam_result.tsv coverage_analysis.tsv.gz riboseq_100AA.tsv.gz 100AA_coordinate.tsv.gz metaT_100AA.tsv.gz rnacode_true_90AA.tsv.xz rnacode_false_90AA.tsv.xz antifam_90AA.tsv metaP_90AA.tsv.gz riboseq_90AA.tsv.gz 90AA_coordinate.tsv.gz metaT_90AA.tsv.gz | GMSC10.100AA.quality.tsv.xz GMSC10.90AA.quality.tsv.xz allpass_100AA.txt allpass_90AA.txt | 
 | 02_statistic.py | Merge all the quality control results | GMSC.cluster.tsv.gz rnacode_true_100AA.tsv.xz rnacode_false_100AA.tsv.xz antifam_result.tsv coverage_analysis.tsv.gz riboseq_100AA.tsv.gz 100AA_coordinate.tsv.gz metaT_100AA.tsv.gz | allquality_100AA.tsv.gz allpass_100AA.txt | 
-| 03_merge_all.py | Merge all the values of quality control results | GMSC10.100AA.quality.tsv.xz 100AA_RNAcode.tsv 100AA_metaT.tsv 100AA_RiboSeq.tsv 100AA_metaP_all.tsv GMSC10.90AA.quality.tsv.xz 90AA_RNAcode.tsv 90AA_metaT.tsv 90AA_RiboSeq.tsv 90AA_metaP.tsv | GMSC10.100AA.quality.tsv.xz GMSC10.90AA.quality.tsv.xz allpass_100AA.txt allpass_90AA.txt GMSC10.100AA.quality_test.tsv GMSC10.90AA.quality_test.tsv | 
+| 03_merge_all.py | Merge all the values of quality control results | GMSC10.100AA.quality.tsv.xz 100AA_RNAcode.tsv 100AA_metaT.tsv 100AA_RiboSeq.tsv 100AA_metaP_all.tsv GMSC10.90AA.quality.tsv.xz 90AA_RNAcode.tsv 90AA_metaT.tsv 90AA_RiboSeq.tsv 90AA_metaP.tsv | GMSC10.100AA.quality.tsv.xz GMSC10.90AA.quality.tsv.xz allpass_100AA.txt allpass_90AA.txt GMSC10.100AA.quality_test.tsv GMSC10.90AA.quality_test.tsv GMSC10.100AA.high_quality.tsv GMSC10.90AA.high_quality.tsv | 

General_Scripts/03_Quality_control/merge_quality_control/03_merge_all.py

@@ -44,6 +44,21 @@ def merge(number,n,infile1,infile2,infile3,infile4,infile5,outfile):
             name = f'GMSC10.{n}AA.{nf[:3]}_{nf[3:6]}_{nf[6:9]}'
             out.write(f'{antifam[name]}\t{terminal[name]}\t{rnacode[name]}\t{metat[name]}\t{riboseq[name]}\t{metap[name]}\n')
 
+def hq(infile,outfile,aa):
+    with open(outfile,'wt') as out:
+        with open(infile,'rt') as f:
+            for n, line in enumerate(f):
+                if line.startswith('AntiFam'):
+                    continue
+                else:
+                    antifam,terminal,rnacode,metat,riboseq,metap = line.strip().split('\t')
+                    if rnacode != 'NA':
+                        if (antifam == 'T' and terminal == 'T' and float(rnacode)<0.05) and (int(metat)>1 or int(riboseq)>1 or round(float(metap),1) >= 0.5):
+                            number = n-1
+                            nf = f'{number:09}'
+                            name = f'GMSC10.{aa}AA.{nf[:3]}_{nf[3:6]}_{nf[6:9]}'
+                            out.write(f'{name}\n')
+
 NUMBER_100 = 964970496
 NUMBER_90 = 287926875
 
@@ -52,13 +67,18 @@ def merge(number,n,infile1,infile2,infile3,infile4,infile5,outfile):
 infile3 = '100AA_metaT.tsv'
 infile4 = '100AA_RiboSeq.tsv'
 infile5 = '100AA_metaP_all.tsv'
-outfile = 'GMSC10.100AA.quality_test.tsv'
-merge(NUMBER_100,100,infile1,infile2,infile3,infile4,infile5,outfile)
+outfile1 = 'GMSC10.100AA.quality_test.tsv'
+merge(NUMBER_100,100,infile1,infile2,infile3,infile4,infile5,outfile1)
 
 infile1 = 'GMSC10.90AA.quality.tsv.xz'
 infile2 = '90AA_RNAcode.tsv'
 infile3 = '90AA_metaT.tsv'
 infile4 = '90AA_RiboSeq.tsv'
 infile5 = '90AA_metaP.tsv'
-outfile = 'GMSC10.90AA.quality_test.tsv'
-merge(NUMBER_90,90,infile1,infile2,infile3,infile4,infile5,outfile)
+outfile2 = 'GMSC10.90AA.quality_test.tsv'
+merge(NUMBER_90,90,infile1,infile2,infile3,infile4,infile5,outfile2)
+
+outfile3 = 'GMSC10.100AA.high_quality.tsv'
+outfile4 = 'GMSC10.90AA.high_quality.tsv'
+hq(outfile1,outfile3,100)
+hq(outfile2,outfile4,90)

General_Scripts/README.md

@@ -85,7 +85,7 @@ The folder contains scripts to generate GMSC resourece from the raw data.
 | :---: | :---: | :---: | :---: |
 | 01_merge.py | Merge all the quality control results | GMSC.cluster.tsv.gz rnacode_true_100AA.tsv.xz rnacode_false_100AA.tsv.xz antifam_result.tsv coverage_analysis.tsv.gz riboseq_100AA.tsv.gz 100AA_coordinate.tsv.gz metaT_100AA.tsv.gz rnacode_true_90AA.tsv.xz rnacode_false_90AA.tsv.xz antifam_90AA.tsv metaP_90AA.tsv.gz riboseq_90AA.tsv.gz 90AA_coordinate.tsv.gz metaT_90AA.tsv.gz | GMSC10.100AA.quality.tsv.xz GMSC10.90AA.quality.tsv.xz allpass_100AA.txt allpass_90AA.txt | 
 | 02_statistic.py | Merge all the quality control results | GMSC.cluster.tsv.gz rnacode_true_100AA.tsv.xz rnacode_false_100AA.tsv.xz antifam_result.tsv coverage_analysis.tsv.gz riboseq_100AA.tsv.gz 100AA_coordinate.tsv.gz metaT_100AA.tsv.gz | allquality_100AA.tsv.gz allpass_100AA.txt | 
-| 03_merge_all.py | Merge all the values of quality control results | GMSC10.100AA.quality.tsv.xz 100AA_RNAcode.tsv 100AA_metaT.tsv 100AA_RiboSeq.tsv 100AA_metaP_all.tsv GMSC10.90AA.quality.tsv.xz 90AA_RNAcode.tsv 90AA_metaT.tsv 90AA_RiboSeq.tsv 90AA_metaP.tsv | GMSC10.100AA.quality.tsv.xz GMSC10.90AA.quality.tsv.xz allpass_100AA.txt allpass_90AA.txt GMSC10.100AA.quality_test.tsv GMSC10.90AA.quality_test.tsv | 
+| 03_merge_all.py | Merge all the values of quality control results | GMSC10.100AA.quality.tsv.xz 100AA_RNAcode.tsv 100AA_metaT.tsv 100AA_RiboSeq.tsv 100AA_metaP_all.tsv GMSC10.90AA.quality.tsv.xz 90AA_RNAcode.tsv 90AA_metaT.tsv 90AA_RiboSeq.tsv 90AA_metaP.tsv | GMSC10.100AA.quality.tsv.xz GMSC10.90AA.quality.tsv.xz allpass_100AA.txt allpass_90AA.txt GMSC10.100AA.quality_test.tsv GMSC10.90AA.quality_test.tsv GMSC10.100AA.high_quality.tsv GMSC10.90AA.high_quality.tsv | 
 
 ## 04_Frozen