@@ -14,7 +14,7 @@ def merge(infile1,infile2,infile3,infile4,infile5,infile6,infile7,infile8,outfil
14
14
with lzma .open (infile1 ,"rt" ) as f1 :
15
15
for line in f1 :
16
16
linelist = line .strip ().split ("\t " )
17
- smorf [linelist [1 ]] = ["NA" ,"T" ,"F" ,"F" ,"NA" ,"F" ]
17
+ smorf [linelist [0 ]] = ["NA" ,"T" ,"F" ,"F" ,"NA" ,"F" ]
18
18
19
19
with lzma .open (infile2 ,"rt" ) as f2 :
20
20
for line in f2 :
@@ -55,22 +55,19 @@ def merge(infile1,infile2,infile3,infile4,infile5,infile6,infile7,infile8,outfil
55
55
smorf [line ][5 ] = "T"
56
56
57
57
for key ,value in smorf .items ():
58
- out .write (key + "\t " + value [0 ]+ "\t " + value [1 ]+ "\t " + value [2 ]+ "\t " + value [3 ]+ "\t " + value [4 ]+ "\t " + value [5 ]+ "\n " )
59
-
58
+ out .write (f'{ key } \t { value [0 ]} \t { value [1 ]} \t { value [2 ]} \t { value [3 ]} \t { value [4 ]} \t { value [5 ]} \n ' )
60
59
out .close ()
61
60
62
61
def allpass (infile ,outfile ):
63
- import gzip
64
-
65
- out = open (outfile ,"wt" )
66
- with gzip .open (infile ,"rt" ) as f1 :
67
- for line in f1 :
68
- linelist = line .strip ().split ("\t " )
69
- if linelist [1 ] == "T" and linelist [2 ] == "T" and linelist [5 ] == "T" and (linelist [3 ] == "T" or linelist [4 ] == "T" or linelist [6 ] == "T" ):
70
- out .write (linelist [0 ]+ "\n " )
71
- out .close ()
72
-
73
- INPUT_FILE_1 = "100AA_rename.tsv.xz"
62
+ import lzma
63
+ with open (outfile ,'wt' ) as out :
64
+ with lzma .open (infile ,"rt" ) as f1 :
65
+ for line in f1 :
66
+ smorf ,rnacode ,antifam ,metap ,riboseq ,terminal ,metat = line .strip ().split ("\t " )
67
+ if rnacode == "T" and antifam == "T" and terminal == "T" and (metap == "T" or riboseq == "T" or metat == "T" ):
68
+ out .write (f'{ smorf } \n ' )
69
+ #100AA
70
+ INPUT_FILE_1 = "GMSC.cluster.tsv.gz"
74
71
INPUT_FILE_2 = "rnacode_true_100AA.tsv"
75
72
INPUT_FILE_3 = "rnacode_false_100AA.tsv"
76
73
INPUT_FILE_4 = "antifam_result.tsv"
@@ -81,5 +78,19 @@ def allpass(infile,outfile):
81
78
OUTPUT_FILE_1 = "GMSC10.100AA.quality.tsv.xz"
82
79
OUTPUT_FILE_2 = "allpass_100AA.txt"
83
80
81
+ merge (INPUT_FILE_1 ,INPUT_FILE_2 ,INPUT_FILE_3 ,INPUT_FILE_4 ,INPUT_FILE_5 ,INPUT_FILE_6 ,INPUT_FILE_7 ,INPUT_FILE_8 ,OUTPUT_FILE_1 )
82
+ allpass (OUTPUT_FILE_1 ,OUTPUT_FILE_2 )
83
+
84
+ INPUT_FILE_1 = "GMSC.cluster.tsv.gz"
85
+ INPUT_FILE_2 = "rnacode_true_90AA.tsv"
86
+ INPUT_FILE_3 = "rnacode_false_90AA.tsv"
87
+ INPUT_FILE_4 = "antifam_90AA.tsv.gz"
88
+ INPUT_FILE_5 = "metaP_90AA.tsv.gz"
89
+ INPUT_FILE_6 = "riboseq_90AA.tsv"
90
+ INPUT_FILE_7 = "90AA_coordinate.tsv.gz"
91
+ INPUT_FILE_8 = "metaT_90AA.tsv"
92
+ OUTPUT_FILE_1 = "GMSC10.90AA.quality.tsv.xz"
93
+ OUTPUT_FILE_2 = "allpass_90AA.txt"
94
+
84
95
merge (INPUT_FILE_1 ,INPUT_FILE_2 ,INPUT_FILE_3 ,INPUT_FILE_4 ,INPUT_FILE_5 ,INPUT_FILE_6 ,INPUT_FILE_7 ,INPUT_FILE_8 ,OUTPUT_FILE_1 )
85
96
allpass (OUTPUT_FILE_1 ,OUTPUT_FILE_2 )
0 commit comments