Skip to content

Commit 8ae61f6

Browse files
robjgiffrobjgiff
committed
Added codon bias scripts.
1 parent 5ff116a commit 8ae61f6

File tree

2 files changed

+364
-0
lines changed

2 files changed

+364
-0
lines changed

‎modules/analysis/refseq/calculateCodonBias.js

Lines changed: 352 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,352 @@
1+
function calculateCodonBias() {
2+
3+
// Script to calculate amino acid composition in reference sequences
4+
5+
var codingFeatures = {};
6+
var resultMap = glue.command(["list", "feature","-w", "featureMetatags.name = 'CODES_AMINO_ACIDS'"]);
7+
var featureList = resultMap["listResult"];
8+
var codingFeatureList = featureList["row"];
9+
_.each(codingFeatureList,function(featureObj){
10+
11+
//glue.log("INFO", "RESULT WAS ", featureObj);
12+
13+
var valueArray = featureObj["value"];
14+
var codingFeatureName = valueArray[0];
15+
//glue.log("INFO", "NAME WAS ", featureName)
16+
codingFeatures[codingFeatureName] = featureObj;
17+
18+
});
19+
20+
//glue.log("INFO", "RESULT WAS ", codingFeatures);
21+
22+
// get list of reference sequences from GLUE
23+
var referencesResult = glue.command(["list","reference"]);
24+
//glue.log("INFO", "RESULT WAS ", referencesResult);
25+
26+
var listResult = referencesResult["listResult"];
27+
var referencesList = listResult["row"];
28+
//glue.log("INFO", "RESULT WAS ", referencesList);
29+
30+
31+
// iterate through reference list and get AA composition of each coding feature
32+
codonCompositionResults = {}
33+
_.each(referencesList, function(refObj) {
34+
35+
//glue.log("INFO", "RESULT WAS ", refObj);
36+
37+
var refseqResults = {};
38+
var referenceProperties = refObj["value"];
39+
var referenceName = referenceProperties[0];
40+
var sequenceID = referenceProperties[2];
41+
42+
//glue.log("INFO", "Reference name result was:", referenceName);
43+
44+
// list all features annotated in this reference
45+
// GLUE COMMAND: reference [referenceName] list feature-location
46+
glue.inMode("/reference/"+referenceName, function() {
47+
48+
var featuresResult = glue.tableToObjects(glue.command(["list", "feature-location"]));
49+
//glue.log("INFO", "RESULT WAS ", featuresResult);
50+
51+
// iterate through features
52+
_.each(featuresResult, function(featureObj) {
53+
54+
//glue.log("INFO", "RESULT WAS ", featureObj);
55+
56+
var featureResults = {};
57+
var aaCounts = {};
58+
59+
// get amino acid sequence
60+
var featureName = featureObj["feature.name"];
61+
//glue.log("INFO", "Feature name result was:", featureName);
62+
63+
// get amino acid sequence of the features
64+
if (codingFeatures[featureName]) {
65+
66+
glue.inMode("/feature-location/"+featureName, function() {
67+
68+
var codonSequenceResult = glue.tableToObjects(glue.command(["amino-acid"]));
69+
70+
//glue.log("INFO", "Table result was:", codonSequenceResult);
71+
72+
// iterate through and get codon composition of feature
73+
var length = 0;
74+
_.each(codonSequenceResult, function(codonObj) {
75+
76+
var codon = codonObj["codonNts"];
77+
var aa = codonObj["aminoAcid"];
78+
79+
// increment amino acid counts
80+
if (aaCounts[aa]) {
81+
82+
aaCounts[aa] += 1;
83+
84+
} else {
85+
aaCounts[aa] = 1;
86+
}
87+
88+
length += 1;
89+
90+
//glue.log("INFO", "Amino acid result:", aa);
91+
//glue.log("INFO", " Codon result:", codon);
92+
93+
if (featureResults[aa]) {
94+
95+
// get the map with codon counts for this aa
96+
aaCodonMap = featureResults[aa];
97+
aaCodonMap[codon] += 1;
98+
99+
}
100+
else {
101+
102+
// create a map with all codons for this aa as keys
103+
aaCodonMap = create_codon_map(aa);
104+
// increment the map for the current codon
105+
106+
aaCodonMap[codon] = 1;
107+
108+
// associate this data structure with the genome feature map
109+
// by using the current aa as a key
110+
featureResults[aa] = aaCodonMap;
111+
112+
}
113+
114+
});
115+
116+
// store the counts for each amino acid
117+
featureResults["aa_counts"] = aaCounts;
118+
119+
// store the total sequence length
120+
featureResults["length"] = length;
121+
122+
// Store the feature name
123+
featureResults["feature_name"] = featureName;
124+
125+
});
126+
127+
//glue.log("INFO", "FEATURE RESULT WAS ", featureResults); die;
128+
129+
// store feature result
130+
refseqResults[featureName] = featureResults;
131+
refseqResults["reference_name"] = referenceName;
132+
refseqResults["sequence_id"] = sequenceID;
133+
134+
}
135+
136+
});
137+
138+
// store reference result
139+
codonCompositionResults[referenceName] = refseqResults;
140+
141+
});
142+
143+
});
144+
145+
// glue.log("INFO", "FINAL RESULT WAS ", codonCompositionResults);
146+
var outputArray = [];
147+
148+
// get each reference
149+
150+
_.each(codonCompositionResults, function(refseqResults) {
151+
152+
//glue.log("INFO", "RESULT WAS ", refseqResults);
153+
154+
var referenceName = refseqResults["reference_name"];
155+
var sequenceID = refseqResults["sequence_id"];
156+
157+
// get each feature
158+
_.each(refseqResults, function(featureResults) {
159+
160+
var featureName = featureResults["feature_name"];
161+
162+
// get each amino
163+
aaCounts = featureResults["aa_counts"];
164+
var referenceName = refseqResults["reference_name"];
165+
_.each(_.keys(aaCounts), function(aa) {
166+
167+
var aaCount = aaCounts[aa];
168+
// glue.log("INFO", "Got count '"+count+"' for aa character'"+aa+"'");
169+
170+
var codonCounts = featureResults[aa];
171+
//glue.log("INFO", "RESULT WAS ", codonCounts); die;
172+
_.each(_.keys(codonCounts), function(codon) {
173+
174+
var codonCount = codonCounts[codon];
175+
176+
// write line: reference - feature - aa - aa count - codon - codon count
177+
outputArray.push({
178+
179+
referenceName: referenceName,
180+
sequenceID: sequenceID,
181+
featureName: featureName,
182+
aa: aa,
183+
aaCount: aaCount,
184+
codon: codon,
185+
codonCount: codonCount
186+
187+
});
188+
189+
190+
});
191+
192+
});
193+
194+
});
195+
196+
});
197+
198+
return outputArray;
199+
}
200+
201+
202+
function create_codon_map(aa){
203+
204+
var codonMap = {};
205+
206+
if (aa == 'A') {
207+
208+
codonMap['GCT'] = 0;
209+
codonMap['GCC'] = 0;
210+
codonMap['GCA'] = 0;
211+
codonMap['GCG'] = 0;
212+
213+
} else if (aa == 'C') {
214+
215+
codonMap['TGT'] = 0;
216+
codonMap['TGC'] = 0;
217+
218+
219+
} else if (aa == 'D') {
220+
221+
codonMap['GAT'] = 0;
222+
codonMap['GAC'] = 0;
223+
224+
} else if (aa == 'E') {
225+
226+
codonMap['GAA'] = 0;
227+
codonMap['GAG'] = 0;
228+
229+
} else if (aa == 'F') {
230+
231+
codonMap['TTT'] = 0;
232+
codonMap['TTC'] = 0;
233+
234+
} else if (aa == 'G') {
235+
236+
codonMap['GGT'] = 0;
237+
codonMap['GGC'] = 0;
238+
codonMap['GGA'] = 0;
239+
codonMap['GGG'] = 0;
240+
241+
} else if (aa == 'H') {
242+
243+
codonMap['CAT'] = 0;
244+
codonMap['CAC'] = 0;
245+
246+
} else if (aa == 'I') {
247+
248+
codonMap['ATT'] = 0;
249+
codonMap['ATC'] = 0;
250+
codonMap['ATA'] = 0;
251+
252+
} else if (aa == 'K') {
253+
254+
codonMap['AAA'] = 0;
255+
codonMap['AAG'] = 0;
256+
257+
} else if (aa == 'L') {
258+
259+
codonMap['CTT'] = 0;
260+
codonMap['CTC'] = 0;
261+
codonMap['CTA'] = 0;
262+
codonMap['CTG'] = 0;
263+
codonMap['TTA'] = 0;
264+
codonMap['TTG'] = 0;
265+
266+
267+
} else if (aa == 'M') {
268+
269+
codonMap['ATG'] = 0;
270+
271+
} else if (aa == 'N') {
272+
273+
codonMap['AAT'] = 0;
274+
codonMap['AAC'] = 0;
275+
276+
} else if (aa == 'P') {
277+
278+
codonMap['CCT'] = 0;
279+
codonMap['CCC'] = 0;
280+
codonMap['CCA'] = 0;
281+
codonMap['CCG'] = 0;
282+
283+
} else if (aa == 'Q') {
284+
285+
codonMap['CAA'] = 0;
286+
codonMap['CAG'] = 0;
287+
288+
} else if (aa == 'R') {
289+
290+
codonMap['CGT'] = 0;
291+
codonMap['CGC'] = 0;
292+
codonMap['CGA'] = 0;
293+
codonMap['CGG'] = 0;
294+
codonMap['AGA'] = 0;
295+
codonMap['AGG'] = 0;
296+
297+
} else if (aa == 'S') {
298+
299+
codonMap['TCT'] = 0;
300+
codonMap['TCC'] = 0;
301+
codonMap['TCA'] = 0;
302+
codonMap['TCG'] = 0;
303+
codonMap['AGT'] = 0;
304+
codonMap['AGC'] = 0;
305+
306+
} else if (aa == 'T') {
307+
308+
codonMap['ACT'] = 0;
309+
codonMap['ACC'] = 0;
310+
codonMap['ACA'] = 0;
311+
codonMap['ACG'] = 0;
312+
313+
} else if (aa == 'V') {
314+
315+
codonMap['GTT'] = 0;
316+
codonMap['GTC'] = 0;
317+
codonMap['GTA'] = 0;
318+
codonMap['GTG'] = 0;
319+
320+
} else if (aa == 'W') {
321+
322+
codonMap['TGG'] = 0;
323+
324+
} else if (aa == 'Y') {
325+
326+
codonMap['TAT'] = 0;
327+
codonMap['TAC'] = 0;
328+
329+
} else if (aa == '*') {
330+
331+
codonMap['TAA'] = 0;
332+
codonMap['TAG'] = 0;
333+
codonMap['TGA'] = 0;
334+
335+
} else if (aa == 'X') {
336+
337+
codonMap['NNN'] = 0;
338+
339+
} else {
340+
341+
glue.log("INFO", "AMINO RESULT WAS ", aa);
342+
die;
343+
344+
345+
}
346+
347+
return codonMap;
348+
349+
}
350+
351+
352+

‎modules/analysis/refseq/calculateCodonBias.xml

Lines changed: 12 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,12 @@
1+
<ecmaFunctionInvoker>
2+
3+
<scriptFileName>calculateCodonBias.js</scriptFileName>
4+
5+
<function>
6+
<name>calculateCodonBias</name>
7+
8+
<tableFromObjectsResultType/>
9+
</function>
10+
11+
</ecmaFunctionInvoker>
12+

0 commit comments

Comments
 (0)