annotate scripts/strike-ambigs.bf @ 34:b110a31a9c61 draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 821e0f3e4546bc55e093ab63bd9f9b57f2bf772d
author iuc
date Thu, 18 Aug 2022 13:53:29 +0000
parents 389fd0e51bcb
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
24
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
1 RequireVersion ("2.5.20");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
2
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
3 LoadFunctionLibrary ("libv3/tasks/alignments.bf");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
4 LoadFunctionLibrary ("libv3/tasks/trees.bf");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
5 LoadFunctionLibrary ("libv3/UtilityFunctions.bf");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
6 LoadFunctionLibrary ("libv3/IOFunctions.bf");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
7 LoadFunctionLibrary ("libv3/convenience/math.bf");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
8
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
9
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
10
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
11 filter.analysis_description = {terms.io.info :
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
12 "
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
13 Read an alignment of coding sequences and replace any ambiguous codons with ---. Write results to a new file in FASTA format, and report changed sequences to stdout
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
14 ",
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
15 terms.io.version : "0.1",
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
16 terms.io.reference : "TBD",
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
17 terms.io.authors : "Sergei L Kosakovsky Pond",
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
18 terms.io.contact : "spond@temple.edu",
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
19 terms.io.requirements : "An MSA"
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
20 };
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
21
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
22
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
23 io.DisplayAnalysisBanner (filter.analysis_description);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
24
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
25 utility.SetEnvVariable ("NORMALIZE_SEQUENCE_NAMES", FALSE);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
26
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
27 KeywordArgument ("code", "Which genetic code should be used", "Universal");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
28 KeywordArgument ("alignment", "An in-frame codon alignment in one of the formats supported by HyPhy");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
29
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
30 filter.in = alignments.PromptForGeneticCodeAndAlignment ("filter.dataset", "filter.input");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
31
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
32 KeywordArgument ("output", ".fasta for compressed data", None);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
33 filter.out = io.PromptUserForFilePath(".fasta for filtered data");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
34 fprintf (filter.out, CLEAR_FILE, KEEP_OPEN);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
35
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
36 GetDataInfo (filter.site_patterns, filter.input);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
37
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
38 filter.patter2site = {};
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
39
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
40
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
41 for (i,j,v; in; filter.site_patterns) {
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
42 index = i+j;
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
43 if (filter.patter2site / v == FALSE ) {
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
44 filter.patter2site [v] = {};
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
45 }
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
46 filter.patter2site [v] + index;
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
47 }
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
48
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
49 GET_DATA_INFO_RETURNS_ONLY_THE_INDEX = TRUE;
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
50 COUNT_GAPS_IN_FREQUENCIES = FALSE;
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
51 filter.unique_patterns = utility.Array1D (filter.input.site_freqs);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
52
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
53 for (seq = 0; seq < filter.input.species; seq += 1) {
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
54 io.ReportProgressBar ("filter","Processing sequence " + (1+seq));
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
55 codons = {1, filter.input.sites};
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
56 codons [0] = "";
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
57 GetString (seq_name, filter.input, seq);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
58 GetDataInfo (seq_chars, filter.input, seq);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
59
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
60 filter.ambigs = 0;
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
61
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
62 for (pattern = 0; pattern < filter.unique_patterns; pattern += 1) {
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
63 GetDataInfo (pattern_info, filter.input, seq, pattern);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
64 if (pattern_info >= 0) {
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
65 codon_start = (filter.patter2site[pattern])[0] * 3;
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
66 codon = seq_chars [codon_start][codon_start+2];
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
67 } else {
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
68 codon = "---";
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
69 filter.ambigs += Abs (filter.patter2site [pattern])
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
70 }
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
71 for (c; in; filter.patter2site [pattern] ) {
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
72 codons[c] = codon;
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
73 }
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
74 }
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
75 if (filter.ambigs > 0) {
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
76 fprintf (stdout, "\nStriking ", filter.ambigs, " codons that are incompletely resolved from " + seq_name + "\n");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
77 }
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
78 fprintf (filter.out,">",seq_name,"\n",Join ("", codons), "\n");
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
79 }
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
80
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
81 fprintf (filter.out,CLOSE_FILE);
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
82
389fd0e51bcb "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/hyphy/ commit 2742ee3b4e90f65352845265d2f85c4263e0eabb"
iuc
parents:
diff changeset
83