msa-parsers 5.0.6 → 5.0.7
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/msa/A3mMSA.js +2 -2
- package/dist/msa/A3mMSA.js.map +1 -1
- package/dist/msa/A3mMSA.test.js +46 -14
- package/dist/msa/A3mMSA.test.js.map +1 -1
- package/dist/msa/FastaMSA.d.ts +1 -0
- package/dist/msa/FastaMSA.js +8 -2
- package/dist/msa/FastaMSA.js.map +1 -1
- package/dist/msa/index.test.js +37 -1
- package/dist/msa/index.test.js.map +1 -1
- package/dist/msa/parseNewick.js +10 -2
- package/dist/msa/parseNewick.js.map +1 -1
- package/dist/src/gff/gffToInterPro.d.ts +20 -0
- package/dist/src/gff/gffToInterPro.js +83 -0
- package/dist/src/gff/gffToInterPro.js.map +1 -0
- package/dist/src/gff/gffToInterPro.test.d.ts +1 -0
- package/dist/src/gff/gffToInterPro.test.js +181 -0
- package/dist/src/gff/gffToInterPro.test.js.map +1 -0
- package/dist/src/gff/index.d.ts +3 -0
- package/dist/src/gff/index.js +4 -0
- package/dist/src/gff/index.js.map +1 -0
- package/dist/src/gff/interProToGFF.d.ts +9 -0
- package/dist/src/gff/interProToGFF.js +48 -0
- package/dist/src/gff/interProToGFF.js.map +1 -0
- package/dist/src/gff/interProToGFF.test.d.ts +1 -0
- package/dist/src/gff/interProToGFF.test.js +189 -0
- package/dist/src/gff/interProToGFF.test.js.map +1 -0
- package/dist/src/gff/parseGFF.d.ts +2 -0
- package/dist/src/gff/parseGFF.js +41 -0
- package/dist/src/gff/parseGFF.js.map +1 -0
- package/dist/src/gff/parseGFF.test.d.ts +1 -0
- package/dist/src/gff/parseGFF.test.js +92 -0
- package/dist/src/gff/parseGFF.test.js.map +1 -0
- package/dist/src/index.d.ts +5 -0
- package/dist/src/index.js +9 -0
- package/dist/src/index.js.map +1 -0
- package/dist/src/msa/A3mMSA.d.ts +33 -0
- package/dist/src/msa/A3mMSA.js +282 -0
- package/dist/src/msa/A3mMSA.js.map +1 -0
- package/dist/src/msa/A3mMSA.test.d.ts +1 -0
- package/dist/src/msa/A3mMSA.test.js +261 -0
- package/dist/src/msa/A3mMSA.test.js.map +1 -0
- package/dist/src/msa/ClustalMSA.d.ts +30 -0
- package/dist/src/msa/ClustalMSA.js +53 -0
- package/dist/src/msa/ClustalMSA.js.map +1 -0
- package/dist/src/msa/EmfMSA.d.ts +27 -0
- package/dist/src/msa/EmfMSA.js +53 -0
- package/dist/src/msa/EmfMSA.js.map +1 -0
- package/dist/src/msa/FastaMSA.d.ts +20 -0
- package/dist/src/msa/FastaMSA.js +75 -0
- package/dist/src/msa/FastaMSA.js.map +1 -0
- package/dist/src/msa/StockholmMSA.d.ts +54 -0
- package/dist/src/msa/StockholmMSA.js +113 -0
- package/dist/src/msa/StockholmMSA.js.map +1 -0
- package/dist/src/msa/index.d.ts +18 -0
- package/dist/src/msa/index.js +34 -0
- package/dist/src/msa/index.js.map +1 -0
- package/dist/src/msa/index.test.d.ts +1 -0
- package/dist/src/msa/index.test.js +96 -0
- package/dist/src/msa/index.test.js.map +1 -0
- package/dist/src/msa/parseNewick.d.ts +60 -0
- package/dist/src/msa/parseNewick.js +105 -0
- package/dist/src/msa/parseNewick.js.map +1 -0
- package/dist/src/msa/stockholmParser.d.ts +22 -0
- package/dist/src/msa/stockholmParser.js +141 -0
- package/dist/src/msa/stockholmParser.js.map +1 -0
- package/dist/src/msa/stockholmParser.test.d.ts +1 -0
- package/dist/src/msa/stockholmParser.test.js +111 -0
- package/dist/src/msa/stockholmParser.test.js.map +1 -0
- package/dist/src/types.d.ts +66 -0
- package/dist/src/types.js +2 -0
- package/dist/src/types.js.map +1 -0
- package/dist/src/util.d.ts +2 -0
- package/dist/src/util.js +10 -0
- package/dist/src/util.js.map +1 -0
- package/package.json +3 -3
- package/src/msa/A3mMSA.test.ts +47 -14
- package/src/msa/A3mMSA.ts +2 -2
- package/src/msa/FastaMSA.ts +14 -3
- package/src/msa/index.test.ts +50 -1
- package/src/msa/parseNewick.ts +14 -2
package/dist/msa/A3mMSA.js
CHANGED
|
@@ -91,7 +91,7 @@ export default class A3mMSA {
|
|
|
91
91
|
if (isLower(code)) {
|
|
92
92
|
hasLowercase = true;
|
|
93
93
|
}
|
|
94
|
-
else if (code >= CODE_A && code <= CODE_Z || code === CODE_DASH) {
|
|
94
|
+
else if ((code >= CODE_A && code <= CODE_Z) || code === CODE_DASH) {
|
|
95
95
|
// Uppercase letters and dashes are match columns
|
|
96
96
|
matchLen++;
|
|
97
97
|
}
|
|
@@ -131,7 +131,7 @@ export default class A3mMSA {
|
|
|
131
131
|
let i = 0;
|
|
132
132
|
while (i < seq.length) {
|
|
133
133
|
const code = seq.charCodeAt(i);
|
|
134
|
-
if (code >= CODE_A && code <= CODE_Z || code === CODE_DASH) {
|
|
134
|
+
if ((code >= CODE_A && code <= CODE_Z) || code === CODE_DASH) {
|
|
135
135
|
// Uppercase letter or dash - match column (dash = deletion)
|
|
136
136
|
matches.push(seq[i]);
|
|
137
137
|
// Collect following lowercase/dot characters as insert content
|
package/dist/msa/A3mMSA.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"A3mMSA.js","sourceRoot":"","sources":["../../src/msa/A3mMSA.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;;;;;;GAoBG;AAEH,sDAAsD;AACtD,MAAM,MAAM,GAAG,EAAE,CAAA,CAAC,MAAM;AACxB,MAAM,MAAM,GAAG,EAAE,CAAA,CAAC,MAAM;AACxB,MAAM,MAAM,GAAG,EAAE,CAAA,CAAC,MAAM;AACxB,MAAM,MAAM,GAAG,GAAG,CAAA,CAAC,MAAM;AACzB,MAAM,SAAS,GAAG,EAAE,CAAA,CAAC,MAAM;AAC3B,MAAM,QAAQ,GAAG,EAAE,CAAA,CAAC,MAAM;AAE1B,SAAS,OAAO,CAAC,IAAY;IAC3B,OAAO,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,CAAA;AACzC,CAAC;AAED,MAAM,CAAC,OAAO,OAAO,MAAM;IACjB,GAAG,CAAqC;IACxC,YAAY,CAAU;IAE9B,YAAY,IAAY;QACtB,MAAM,OAAO,GAAa,EAAE,CAAA;QAC5B,MAAM,KAAK,GAAa,EAAE,CAAA;QAE1B,6DAA6D;QAC7D,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YACpC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,UAAU,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;YACtC,IAAI,UAAU,KAAK,CAAC,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,OAAO,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,UAAU,CAAC,CAAA;YAC1C,MAAM,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAA;YACrC,MAAM,EAAE,GAAG,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAA;YACjE,IAAI,EAAE,EAAE,CAAC;gBACP,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,UAAU,GAAG,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,CAAA;gBAC/D,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YAChB,CAAC;QACH,CAAC;QAED,IAAI,CAAC,YAAY,GAAG,KAAK,CAAA;QACzB,IAAI,CAAC,GAAG,GAAG,EAAE,OAAO,EAAE,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,KAAK,CAAC,EAAE,CAAA;IACxD,CAAC;IAED;;OAEG;IACH,MAAM,CAAC,KAAK,CAAC,IAAY;QACvB,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YAC1B,OAAO,KAAK,CAAA;QACd,CAAC;QAED,MAAM,IAAI,GAAa,EAAE,CAAA;QACzB,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YACpC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,UAAU,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;YACtC,IAAI,UAAU,KAAK,CAAC,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,GAAG,GAAG,KAAK,CAAC,KAAK,CAAC,UAAU,GAAG,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,EAAE,EAAE,CAAC,CAAA;YAC7D,IAAI,GAAG,EAAE,CAAC;gBACR,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;YAChB,CAAC;QACH,CAAC;QAED,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACpB,OAAO,KAAK,CAAA;QACd,CAAC;QAED,mFAAmF;QACnF,iEAAiE;QACjE,IAAI,YAAY,GAAG,KAAK,CAAA;QACxB,IAAI,aAAa,GAAG,CAAC,CAAC,CAAA;QACtB,IAAI,eAAe,GAAG,IAAI,CAAA;QAE1B,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACvB,IAAI,QAAQ,GAAG,CAAC,CAAA;YAChB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBACpC,MAAM,IAAI,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAA;gBAC9B,IAAI,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;oBAClB,YAAY,GAAG,IAAI,CAAA;gBACrB,CAAC;qBAAM,IAAI,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;
|
|
1
|
+
{"version":3,"file":"A3mMSA.js","sourceRoot":"","sources":["../../src/msa/A3mMSA.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;;;;;;GAoBG;AAEH,sDAAsD;AACtD,MAAM,MAAM,GAAG,EAAE,CAAA,CAAC,MAAM;AACxB,MAAM,MAAM,GAAG,EAAE,CAAA,CAAC,MAAM;AACxB,MAAM,MAAM,GAAG,EAAE,CAAA,CAAC,MAAM;AACxB,MAAM,MAAM,GAAG,GAAG,CAAA,CAAC,MAAM;AACzB,MAAM,SAAS,GAAG,EAAE,CAAA,CAAC,MAAM;AAC3B,MAAM,QAAQ,GAAG,EAAE,CAAA,CAAC,MAAM;AAE1B,SAAS,OAAO,CAAC,IAAY;IAC3B,OAAO,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,CAAA;AACzC,CAAC;AAED,MAAM,CAAC,OAAO,OAAO,MAAM;IACjB,GAAG,CAAqC;IACxC,YAAY,CAAU;IAE9B,YAAY,IAAY;QACtB,MAAM,OAAO,GAAa,EAAE,CAAA;QAC5B,MAAM,KAAK,GAAa,EAAE,CAAA;QAE1B,6DAA6D;QAC7D,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YACpC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,UAAU,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;YACtC,IAAI,UAAU,KAAK,CAAC,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,OAAO,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,UAAU,CAAC,CAAA;YAC1C,MAAM,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAA;YACrC,MAAM,EAAE,GAAG,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAA;YACjE,IAAI,EAAE,EAAE,CAAC;gBACP,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,UAAU,GAAG,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,CAAA;gBAC/D,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YAChB,CAAC;QACH,CAAC;QAED,IAAI,CAAC,YAAY,GAAG,KAAK,CAAA;QACzB,IAAI,CAAC,GAAG,GAAG,EAAE,OAAO,EAAE,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,KAAK,CAAC,EAAE,CAAA;IACxD,CAAC;IAED;;OAEG;IACH,MAAM,CAAC,KAAK,CAAC,IAAY;QACvB,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YAC1B,OAAO,KAAK,CAAA;QACd,CAAC;QAED,MAAM,IAAI,GAAa,EAAE,CAAA;QACzB,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YACpC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,UAAU,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;YACtC,IAAI,UAAU,KAAK,CAAC,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,GAAG,GAAG,KAAK,CAAC,KAAK,CAAC,UAAU,GAAG,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,EAAE,EAAE,CAAC,CAAA;YAC7D,IAAI,GAAG,EAAE,CAAC;gBACR,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;YAChB,CAAC;QACH,CAAC;QAED,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACpB,OAAO,KAAK,CAAA;QACd,CAAC;QAED,mFAAmF;QACnF,iEAAiE;QACjE,IAAI,YAAY,GAAG,KAAK,CAAA;QACxB,IAAI,aAAa,GAAG,CAAC,CAAC,CAAA;QACtB,IAAI,eAAe,GAAG,IAAI,CAAA;QAE1B,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACvB,IAAI,QAAQ,GAAG,CAAC,CAAA;YAChB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBACpC,MAAM,IAAI,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAA;gBAC9B,IAAI,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;oBAClB,YAAY,GAAG,IAAI,CAAA;gBACrB,CAAC;qBAAM,IAAI,CAAC,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;oBACpE,iDAAiD;oBACjD,QAAQ,EAAE,CAAA;gBACZ,CAAC;YACH,CAAC;YAED,IAAI,aAAa,KAAK,CAAC,CAAC,EAAE,CAAC;gBACzB,aAAa,GAAG,QAAQ,CAAA;YAC1B,CAAC;iBAAM,CAAC;gBACN,IAAI,QAAQ,KAAK,aAAa,EAAE,CAAC;oBAC/B,eAAe,GAAG,KAAK,CAAA;gBACzB,CAAC;YACH,CAAC;QACH,CAAC;QAED,OAAO,YAAY,IAAI,eAAe,CAAA;IACxC,CAAC;IAED;;;;;;;OAOG;IACK,SAAS,CACf,OAAiB,EACjB,KAAe;QAEf,MAAM,OAAO,GAAG,KAAK,CAAC,MAAM,CAAA;QAC5B,IAAI,OAAO,KAAK,CAAC,EAAE,CAAC;YAClB,OAAO,EAAE,CAAA;QACX,CAAC;QAED,2EAA2E;QAC3E,yEAAyE;QACzE,MAAM,UAAU,GAAe,EAAE,CAAA;QACjC,MAAM,aAAa,GAAe,EAAE,CAAA;QAEpC,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,OAAO,EAAE,MAAM,EAAE,EAAE,CAAC;YAChD,MAAM,GAAG,GAAG,OAAO,CAAC,MAAM,CAAE,CAAA;YAC5B,MAAM,OAAO,GAAa,EAAE,CAAA;YAC5B,MAAM,OAAO,GAAa,EAAE,CAAA;YAC5B,IAAI,CAAC,GAAG,CAAC,CAAA;YAET,OAAO,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC;gBACtB,MAAM,IAAI,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAA;gBAE9B,IAAI,CAAC,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;oBAC7D,4DAA4D;oBAC5D,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAE,CAAC,CAAA;oBACrB,+DAA+D;oBAC/D,wDAAwD;oBACxD,IAAI,GAAG,GAAG,EAAE,CAAA;oBACZ,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;oBACb,OAAO,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC;wBACtB,MAAM,CAAC,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAA;wBAC3B,IAAI,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,QAAQ,EAAE,CAAC;4BACjC,GAAG,IAAI,GAAG,CAAC,CAAC,CAAE,CAAA;4BACd,CAAC,EAAE,CAAA;wBACL,CAAC;6BAAM,CAAC;4BACN,MAAK;wBACP,CAAC;oBACH,CAAC;oBACD,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;oBACjB,CAAC,GAAG,CAAC,CAAA;gBACP,CAAC;qBAAM,IAAI,IAAI,KAAK,QAAQ,EAAE,CAAC;oBAC7B,6CAA6C;oBAC7C,CAAC,EAAE,CAAA;gBACL,CAAC;qBAAM,IAAI,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;oBACzB,oCAAoC;oBACpC,IAAI,GAAG,GAAG,EAAE,CAAA;oBACZ,OAAO,CAAC,GAAG,GAAG,CAAC,MAAM,IAAI,OAAO,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;wBACpD,GAAG,IAAI,GAAG,CAAC,CAAC,CAAE,CAAA;wBACd,CAAC,EAAE,CAAA;oBACL,CAAC;oBACD,mCAAmC;oBACnC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;oBAChB,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;gBACnB,CAAC;qBAAM,CAAC;oBACN,CAAC,EAAE,CAAA;gBACL,CAAC;YACH,CAAC;YAED,UAAU,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;YACxB,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;QAC7B,CAAC;QAED,oEAAoE;QACpE,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAA;QAElE,2EAA2E;QAC3E,MAAM,UAAU,GAAG,IAAI,KAAK,CAAS,YAAY,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;QAC1D,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,OAAO,EAAE,MAAM,EAAE,EAAE,CAAC;YAChD,MAAM,OAAO,GAAG,aAAa,CAAC,MAAM,CAAE,CAAA;YACtC,KAAK,IAAI,GAAG,GAAG,CAAC,EAAE,GAAG,GAAG,OAAO,CAAC,MAAM,EAAE,GAAG,EAAE,EAAE,CAAC;gBAC9C,oDAAoD;gBACpD,IAAI,OAAO,GAAG,CAAC,CAAA;gBACf,KAAK,MAAM,CAAC,IAAI,OAAO,CAAC,GAAG,CAAE,EAAE,CAAC;oBAC9B,IAAI,OAAO,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;wBAC7B,OAAO,EAAE,CAAA;oBACX,CAAC;gBACH,CAAC;gBACD,IAAI,OAAO,GAAG,UAAU,CAAC,GAAG,CAAE,EAAE,CAAC;oBAC/B,UAAU,CAAC,GAAG,CAAC,GAAG,OAAO,CAAA;gBAC3B,CAAC;YACH,CAAC;QACH,CAAC;QAED,2BAA2B;QAC3B,MAAM,QAAQ,GAA2B,EAAE,CAAA;QAE3C,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,OAAO,EAAE,MAAM,EAAE,EAAE,CAAC;YAChD,MAAM,OAAO,GAAG,UAAU,CAAC,MAAM,CAAE,CAAA;YACnC,MAAM,OAAO,GAAG,aAAa,CAAC,MAAM,CAAE,CAAA;YACtC,MAAM,MAAM,GAAa,EAAE,CAAA;YAE3B,KAAK,IAAI,GAAG,GAAG,CAAC,EAAE,GAAG,GAAG,YAAY,EAAE,GAAG,EAAE,EAAE,CAAC;gBAC5C,MAAM,MAAM,GAAG,UAAU,CAAC,GAAG,CAAE,CAAA;gBAE/B,IAAI,GAAG,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC;oBACzB,MAAM,SAAS,GAAG,OAAO,CAAC,GAAG,CAAE,CAAA;oBAC/B,MAAM,UAAU,GAAG,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,CAAA;oBAErC,wCAAwC;oBACxC,MAAM,CAAC,IAAI,CAAC,SAAS,IAAI,GAAG,CAAC,CAAA;oBAE7B,yBAAyB;oBACzB,IAAI,SAAS,GAAG,EAAE,CAAA;oBAClB,KAAK,MAAM,CAAC,IAAI,UAAU,EAAE,CAAC;wBAC3B,IAAI,OAAO,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;4BAC7B,SAAS,IAAI,CAAC,CAAC,WAAW,EAAE,CAAA;wBAC9B,CAAC;oBACH,CAAC;oBAED,sCAAsC;oBACtC,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;oBAEtB,2CAA2C;oBAC3C,MAAM,OAAO,GAAG,MAAM,GAAG,SAAS,CAAC,MAAM,CAAA;oBACzC,IAAI,OAAO,GAAG,CAAC,EAAE,CAAC;wBAChB,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAA;oBAClC,CAAC;gBACH,CAAC;qBAAM,CAAC;oBACN,sCAAsC;oBACtC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;oBAChB,IAAI,MAAM,GAAG,CAAC,EAAE,CAAC;wBACf,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAA;oBACjC,CAAC;gBACH,CAAC;YACH,CAAC;YAED,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAE,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;QAC5C,CAAC;QAED,OAAO,QAAQ,CAAA;IACjB,CAAC;IAED,MAAM;QACJ,OAAO,IAAI,CAAC,GAAG,CAAA;IACjB,CAAC;IAED,UAAU;QACR,OAAO,SAAS,CAAA;IAClB,CAAC;IAED,QAAQ;QACN,OAAO,IAAI,CAAC,YAAY,CAAA;IAC1B,CAAC;IAED,MAAM,CAAC,IAAY;QACjB,OAAO,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,EAAE,CAAA;IACrC,CAAC;IAED,QAAQ;QACN,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAA;QAC7C,OAAO,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAA;IAC5C,CAAC;IAED,aAAa;QACX,OAAO,EAAE,CAAA;IACX,CAAC;IAED,IAAI,cAAc;QAChB,OAAO,EAAE,CAAA;IACX,CAAC;IAED,SAAS;QACP,OAAO,EAAE,CAAA;IACX,CAAC;IAED,OAAO;QACL,OAAO;YACL,EAAE,EAAE,MAAM;YACV,IAAI,EAAE,MAAM;YACZ,MAAM,EAAE,IAAI;YACZ,QAAQ,EAAE,IAAI,CAAC,QAAQ,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;gBACrC,EAAE,EAAE,IAAI;gBACR,QAAQ,EAAE,EAAE;gBACZ,IAAI;aACL,CAAC,CAAC;SACJ,CAAA;IACH,CAAC;IAED,IAAI,YAAY;QACd,OAAO,SAAS,CAAA;IAClB,CAAC;IAED,IAAI,2BAA2B;QAC7B,OAAO,SAAS,CAAA;IAClB,CAAC;IAED,IAAI,MAAM;QACR,OAAO,EAAE,CAAA;IACX,CAAC;CACF"}
|
package/dist/msa/A3mMSA.test.js
CHANGED
|
@@ -1,5 +1,17 @@
|
|
|
1
1
|
import { describe, expect, test } from 'vitest';
|
|
2
2
|
import A3mMSA from "./A3mMSA.js";
|
|
3
|
+
// A3M format rules (see https://yanglab.qd.sdu.edu.cn/trRosetta/msa_format.html):
|
|
4
|
+
//
|
|
5
|
+
// - Uppercase letters = "match columns" (aligned positions shared by all seqs)
|
|
6
|
+
// - Lowercase letters = insertions (extra residues in one seq, not a match column)
|
|
7
|
+
// - Dashes '-' = deletions (missing residue at a match column, counts as match)
|
|
8
|
+
// - Dots '.' = gaps aligned to insertions in other sequences
|
|
9
|
+
//
|
|
10
|
+
// Key constraint: all sequences must have the same number of match columns
|
|
11
|
+
// (uppercase + dashes). Lowercase chars don't count toward this total.
|
|
12
|
+
//
|
|
13
|
+
// When expanded to a standard alignment, lowercase inserts become new columns
|
|
14
|
+
// (uppercased), and sequences without inserts at that position get '.' gaps.
|
|
3
15
|
describe('A3mMSA', () => {
|
|
4
16
|
describe('sniff', () => {
|
|
5
17
|
test('returns false for non-FASTA text', () => {
|
|
@@ -14,10 +26,13 @@ ACDEFGHIKLMNPQRSTVWY`;
|
|
|
14
26
|
expect(A3mMSA.sniff(fasta)).toBe(false);
|
|
15
27
|
});
|
|
16
28
|
test('returns true for A3M format', () => {
|
|
29
|
+
// seq1: match=ACDEFKLMNPQ (11), insert=ghi (3)
|
|
30
|
+
// seq2: match=ACDEFKLMNPQ (11)
|
|
31
|
+
// Same match count + has lowercase → detected as A3M
|
|
17
32
|
const a3m = `>seq1
|
|
18
33
|
ACDEFghiKLMNPQ
|
|
19
34
|
>seq2
|
|
20
|
-
|
|
35
|
+
ACDEFKLMNPQ`;
|
|
21
36
|
expect(A3mMSA.sniff(a3m)).toBe(true);
|
|
22
37
|
});
|
|
23
38
|
test('returns false for single sequence', () => {
|
|
@@ -26,11 +41,13 @@ ACDEFghiKLMNPQ`;
|
|
|
26
41
|
expect(A3mMSA.sniff(a3m)).toBe(false);
|
|
27
42
|
});
|
|
28
43
|
test('counts dashes as match columns in sniff', () => {
|
|
29
|
-
//
|
|
44
|
+
// seq1: match=ACDEF (5), insert=ghi (3)
|
|
45
|
+
// seq2: match=AC--F (5, where -- are deletions that count as match columns)
|
|
46
|
+
// Same match count + has lowercase → detected as A3M
|
|
30
47
|
const a3m = `>seq1
|
|
31
48
|
ACDEFghi
|
|
32
49
|
>seq2
|
|
33
|
-
--
|
|
50
|
+
AC--F`;
|
|
34
51
|
expect(A3mMSA.sniff(a3m)).toBe(true);
|
|
35
52
|
});
|
|
36
53
|
test('returns false when match column counts differ', () => {
|
|
@@ -46,10 +63,14 @@ ACDEF---`;
|
|
|
46
63
|
});
|
|
47
64
|
describe('parsing', () => {
|
|
48
65
|
test('parses simple A3M', () => {
|
|
66
|
+
// Both have 11 match columns. seq1 has 3 lowercase inserts (ghi) after F.
|
|
67
|
+
// After expansion, ghi becomes 3 new columns (uppercased in seq1, dots in seq2):
|
|
68
|
+
// seq1: ACDEF + GHI + KLMNPQ = ACDEFGHIKLMNPQ (14 chars)
|
|
69
|
+
// seq2: ACDEF + ... + KLMNPQ = ACDEF...KLMNPQ (14 chars)
|
|
49
70
|
const a3m = `>seq1
|
|
50
71
|
ACDEFghiKLMNPQ
|
|
51
72
|
>seq2
|
|
52
|
-
|
|
73
|
+
ACDEFKLMNPQ`;
|
|
53
74
|
const msa = new A3mMSA(a3m);
|
|
54
75
|
expect(msa.getNames()).toEqual(['seq1', 'seq2']);
|
|
55
76
|
const seq1 = msa.getRow('seq1');
|
|
@@ -145,26 +166,37 @@ A-FG`;
|
|
|
145
166
|
expect(seq1).toBe('A-..FG');
|
|
146
167
|
});
|
|
147
168
|
test('realistic a3m with insertions and deletions', () => {
|
|
148
|
-
//
|
|
169
|
+
// All 3 sequences have 11 match columns (uppercase + dashes).
|
|
170
|
+
// seq1 has 3 lowercase insertions (ghi) after the 5th match column.
|
|
171
|
+
// seq2 uses dashes for deletions at match positions 1-2 and 9-10.
|
|
172
|
+
//
|
|
173
|
+
// A3M input (match columns marked with ^, inserts unmarked):
|
|
174
|
+
// query: A C D E F K L M N P Q
|
|
175
|
+
// ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ = 11 match cols
|
|
176
|
+
// seq1: A C D E F g h i K L M N P Q
|
|
177
|
+
// ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ = 11 match cols, ghi = inserts
|
|
178
|
+
// seq2: - - D E F K L M - - Q
|
|
179
|
+
// ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ = 11 match cols (dashes are deletions)
|
|
180
|
+
//
|
|
181
|
+
// After expansion (inserts become new columns, '.' fills gaps):
|
|
182
|
+
// query: ACDEF...KLMNPQ (14 chars)
|
|
183
|
+
// seq1: ACDEFGHIKLMNPQ (14 chars, ghi uppercased)
|
|
184
|
+
// seq2: --DEF...KLM--Q (14 chars)
|
|
149
185
|
const a3m = `>query
|
|
150
|
-
|
|
186
|
+
ACDEFKLMNPQ
|
|
151
187
|
>seq1
|
|
152
188
|
ACDEFghiKLMNPQ
|
|
153
189
|
>seq2
|
|
154
|
-
--
|
|
190
|
+
--DEFKLM--Q`;
|
|
155
191
|
const msa = new A3mMSA(a3m);
|
|
156
192
|
const query = msa.getRow('query');
|
|
157
193
|
const seq1 = msa.getRow('seq1');
|
|
158
194
|
const seq2 = msa.getRow('seq2');
|
|
159
195
|
expect(query.length).toBe(seq1.length);
|
|
160
196
|
expect(query.length).toBe(seq2.length);
|
|
161
|
-
|
|
162
|
-
|
|
163
|
-
|
|
164
|
-
// seq2: --DEF + ... + ---KLMNPQ = --DEF...---KLMNPQ
|
|
165
|
-
expect(query).toBe('ACDEF...GHIKLMNPQ');
|
|
166
|
-
expect(seq1).toBe('ACDEFGHIGHIKLMNPQ');
|
|
167
|
-
expect(seq2).toBe('--DEF...---KLMNPQ');
|
|
197
|
+
expect(query).toBe('ACDEF...KLMNPQ');
|
|
198
|
+
expect(seq1).toBe('ACDEFGHIKLMNPQ');
|
|
199
|
+
expect(seq2).toBe('--DEF...KLM--Q');
|
|
168
200
|
});
|
|
169
201
|
test('getWidth returns correct width', () => {
|
|
170
202
|
const a3m = `>seq1
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"A3mMSA.test.js","sourceRoot":"","sources":["../../src/msa/A3mMSA.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,QAAQ,CAAA;AAE/C,OAAO,MAAM,MAAM,aAAa,CAAA;AAEhC,QAAQ,CAAC,QAAQ,EAAE,GAAG,EAAE;IACtB,QAAQ,CAAC,OAAO,EAAE,GAAG,EAAE;QACrB,IAAI,CAAC,kCAAkC,EAAE,GAAG,EAAE;YAC5C,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;YAC7C,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAC/C,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,iCAAiC,EAAE,GAAG,EAAE;YAC3C,MAAM,KAAK,GAAG;;;qBAGC,CAAA;YACf,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QACzC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,6BAA6B,EAAE,GAAG,EAAE;YACvC,MAAM,GAAG,GAAG;;;
|
|
1
|
+
{"version":3,"file":"A3mMSA.test.js","sourceRoot":"","sources":["../../src/msa/A3mMSA.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,QAAQ,CAAA;AAE/C,OAAO,MAAM,MAAM,aAAa,CAAA;AAEhC,kFAAkF;AAClF,EAAE;AACF,+EAA+E;AAC/E,mFAAmF;AACnF,uFAAuF;AACvF,sEAAsE;AACtE,EAAE;AACF,2EAA2E;AAC3E,uEAAuE;AACvE,EAAE;AACF,8EAA8E;AAC9E,6EAA6E;AAE7E,QAAQ,CAAC,QAAQ,EAAE,GAAG,EAAE;IACtB,QAAQ,CAAC,OAAO,EAAE,GAAG,EAAE;QACrB,IAAI,CAAC,kCAAkC,EAAE,GAAG,EAAE;YAC5C,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;YAC7C,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAC/C,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,iCAAiC,EAAE,GAAG,EAAE;YAC3C,MAAM,KAAK,GAAG;;;qBAGC,CAAA;YACf,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QACzC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,6BAA6B,EAAE,GAAG,EAAE;YACvC,+CAA+C;YAC/C,+BAA+B;YAC/B,qDAAqD;YACrD,MAAM,GAAG,GAAG;;;YAGN,CAAA;YACN,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACtC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,mCAAmC,EAAE,GAAG,EAAE;YAC7C,MAAM,GAAG,GAAG;eACH,CAAA;YACT,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QACvC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,yCAAyC,EAAE,GAAG,EAAE;YACnD,wCAAwC;YACxC,4EAA4E;YAC5E,qDAAqD;YACrD,MAAM,GAAG,GAAG;;;MAGZ,CAAA;YACA,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACtC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,+CAA+C,EAAE,GAAG,EAAE;YACzD,gCAAgC;YAChC,mCAAmC;YACnC,8DAA8D;YAC9D,MAAM,MAAM,GAAG;;;SAGZ,CAAA;YACH,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAC1C,CAAC,CAAC,CAAA;IACJ,CAAC,CAAC,CAAA;IAEF,QAAQ,CAAC,SAAS,EAAE,GAAG,EAAE;QACvB,IAAI,CAAC,mBAAmB,EAAE,GAAG,EAAE;YAC7B,0EAA0E;YAC1E,iFAAiF;YACjF,2DAA2D;YAC3D,2DAA2D;YAC3D,MAAM,GAAG,GAAG;;;YAGN,CAAA;YACN,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;YAChD,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAC/B,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAE/B,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YACrC,MAAM,CAAC,IAAI,CAAC,CAAC,SAAS,CAAC,KAAK,CAAC,CAAA;QAC/B,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,8BAA8B,EAAE,GAAG,EAAE;YACxC,iEAAiE;YACjE,kDAAkD;YAClD,MAAM,GAAG,GAAG;;;GAGf,CAAA;YACG,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAC/B,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAE/B,iDAAiD;YACjD,uDAAuD;YACvD,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;YAC1B,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;QAC5B,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,6BAA6B,EAAE,GAAG,EAAE;YACvC,6DAA6D;YAC7D,kDAAkD;YAClD,MAAM,GAAG,GAAG;;;IAGd,CAAA;YACE,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAC/B,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAE/B,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YACrC,0CAA0C;YAC1C,0CAA0C;YAC1C,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,CAAA;YAC9B,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,CAAA;QAChC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,4CAA4C,EAAE,GAAG,EAAE;YACtD,kCAAkC;YAClC,iEAAiE;YACjE,MAAM,GAAG,GAAG;;;OAGX,CAAA;YACD,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;YACjC,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAE/B,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YACtC,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YAC5B,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;QAC7B,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,kCAAkC,EAAE,GAAG,EAAE;YAC5C,yBAAyB;YACzB,gDAAgD;YAChD,MAAM,GAAG,GAAG;;;OAGX,CAAA;YACD,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;YACjC,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAE/B,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YACtC,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YAC5B,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;QAC7B,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,mCAAmC,EAAE,GAAG,EAAE;YAC7C,MAAM,GAAG,GAAG;;;OAGX,CAAA;YACD,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;YACjC,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAE/B,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YACtC,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YAC5B,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;QAC7B,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,gCAAgC,EAAE,GAAG,EAAE;YAC1C,yDAAyD;YACzD,oDAAoD;YACpD,MAAM,GAAG,GAAG;;;KAGb,CAAA;YACC,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;YACjC,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAE/B,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YACtC,qCAAqC;YACrC,oCAAoC;YACpC,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YAC5B,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;QAC7B,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,6CAA6C,EAAE,GAAG,EAAE;YACvD,8DAA8D;YAC9D,oEAAoE;YACpE,kEAAkE;YAClE,EAAE;YACF,6DAA6D;YAC7D,iCAAiC;YACjC,kDAAkD;YAClD,uCAAuC;YACvC,uEAAuE;YACvE,iCAAiC;YACjC,yEAAyE;YACzE,EAAE;YACF,gEAAgE;YAChE,sCAAsC;YACtC,sDAAsD;YACtD,sCAAsC;YACtC,MAAM,GAAG,GAAG;;;;;YAKN,CAAA;YACN,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;YACjC,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAC/B,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAE/B,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YACtC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YAEtC,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YACpC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YACnC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;QACrC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,gCAAgC,EAAE,GAAG,EAAE;YAC1C,MAAM,GAAG,GAAG;;;MAGZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;QAChC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,wBAAwB,EAAE,GAAG,EAAE;YAClC,MAAM,GAAG,GAAG;;;MAGZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAC3B,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,EAAE,CAAA;YAEzB,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,cAAc,CAAC,MAAM,CAAC,CAAA;YAC3C,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,cAAc,CAAC,MAAM,CAAC,CAAA;QAC7C,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,kCAAkC,EAAE,GAAG,EAAE;YAC5C,MAAM,GAAG,GAAG;;;MAGZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAC3B,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,EAAE,CAAA;YAE1B,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAA;QACvC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,yBAAyB,EAAE,GAAG,EAAE;YACnC,MAAM,GAAG,GAAG;MACZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,CAAA;QAC1C,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,2CAA2C,EAAE,GAAG,EAAE;YACrD,MAAM,GAAG,GAAG;;;MAGZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;QAClD,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,0BAA0B,EAAE,GAAG,EAAE;YACpC,MAAM,GAAG,GAAG;;;;;MAKZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,OAAO,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC,CAAA;QAC7D,CAAC,CAAC,CAAA;IACJ,CAAC,CAAC,CAAA;IAEF,QAAQ,CAAC,YAAY,EAAE,GAAG,EAAE;QAC1B,IAAI,CAAC,+BAA+B,EAAE,GAAG,EAAE;YACzC,MAAM,GAAG,GAAG;MACZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,CAAA;QACxC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,2BAA2B,EAAE,GAAG,EAAE;YACrC,MAAM,GAAG,GAAG;MACZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC,aAAa,EAAE,CAAA;QAC1C,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,0CAA0C,EAAE,GAAG,EAAE;YACpD,MAAM,GAAG,GAAG;MACZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,2BAA2B,CAAC,CAAC,aAAa,EAAE,CAAA;QACzD,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,uBAAuB,EAAE,GAAG,EAAE;YACjC,MAAM,GAAG,GAAG;MACZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,CAAA;QAChC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,oCAAoC,EAAE,GAAG,EAAE;YAC9C,MAAM,GAAG,GAAG;MACZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,aAAa,EAAE,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,CAAA;QACzC,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,gCAAgC,EAAE,GAAG,EAAE;YAC1C,MAAM,GAAG,GAAG;MACZ,CAAA;YACA,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,CAAA;YAE3B,MAAM,CAAC,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,CAAA;QACrC,CAAC,CAAC,CAAA;IACJ,CAAC,CAAC,CAAA;AACJ,CAAC,CAAC,CAAA"}
|
package/dist/msa/FastaMSA.d.ts
CHANGED
package/dist/msa/FastaMSA.js
CHANGED
|
@@ -2,6 +2,7 @@ export default class FastaMSA {
|
|
|
2
2
|
MSA;
|
|
3
3
|
constructor(text) {
|
|
4
4
|
const seqdata = {};
|
|
5
|
+
const colonNormalized = {};
|
|
5
6
|
for (const entry of text.split('>')) {
|
|
6
7
|
if (!/\S/.test(entry)) {
|
|
7
8
|
continue;
|
|
@@ -15,9 +16,12 @@ export default class FastaMSA {
|
|
|
15
16
|
const id = spaceIdx === -1 ? defLine : defLine.slice(0, spaceIdx);
|
|
16
17
|
if (id) {
|
|
17
18
|
seqdata[id] = entry.slice(newlineIdx + 1).replaceAll(/\s/g, '');
|
|
19
|
+
if (id.includes(':')) {
|
|
20
|
+
colonNormalized[id.replaceAll(':', '_')] = id;
|
|
21
|
+
}
|
|
18
22
|
}
|
|
19
23
|
}
|
|
20
|
-
this.MSA = { seqdata };
|
|
24
|
+
this.MSA = { seqdata, colonNormalized };
|
|
21
25
|
}
|
|
22
26
|
getMSA() {
|
|
23
27
|
return this.MSA;
|
|
@@ -29,7 +33,9 @@ export default class FastaMSA {
|
|
|
29
33
|
return Object.keys(this.MSA.seqdata);
|
|
30
34
|
}
|
|
31
35
|
getRow(name) {
|
|
32
|
-
return this.MSA.seqdata[name]
|
|
36
|
+
return (this.MSA.seqdata[name] ??
|
|
37
|
+
this.MSA.seqdata[this.MSA.colonNormalized[name] ?? ''] ??
|
|
38
|
+
'');
|
|
33
39
|
}
|
|
34
40
|
getWidth() {
|
|
35
41
|
const name = Object.keys(this.MSA.seqdata)[0];
|
package/dist/msa/FastaMSA.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"FastaMSA.js","sourceRoot":"","sources":["../../src/msa/FastaMSA.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,OAAO,OAAO,QAAQ;IACnB,GAAG,
|
|
1
|
+
{"version":3,"file":"FastaMSA.js","sourceRoot":"","sources":["../../src/msa/FastaMSA.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,OAAO,OAAO,QAAQ;IACnB,GAAG,CAGV;IAED,YAAY,IAAY;QACtB,MAAM,OAAO,GAA2B,EAAE,CAAA;QAC1C,MAAM,eAAe,GAA2B,EAAE,CAAA;QAClD,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YACpC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,UAAU,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;YACtC,IAAI,UAAU,KAAK,CAAC,CAAC,EAAE,CAAC;gBACtB,SAAQ;YACV,CAAC;YACD,MAAM,OAAO,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,UAAU,CAAC,CAAA;YAC1C,MAAM,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAA;YACrC,MAAM,EAAE,GAAG,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAA;YACjE,IAAI,EAAE,EAAE,CAAC;gBACP,OAAO,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC,UAAU,GAAG,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,EAAE,EAAE,CAAC,CAAA;gBAC/D,IAAI,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;oBACrB,eAAe,CAAC,EAAE,CAAC,UAAU,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,GAAG,EAAE,CAAA;gBAC/C,CAAC;YACH,CAAC;QACH,CAAC;QACD,IAAI,CAAC,GAAG,GAAG,EAAE,OAAO,EAAE,eAAe,EAAE,CAAA;IACzC,CAAC;IAED,MAAM;QACJ,OAAO,IAAI,CAAC,GAAG,CAAA;IACjB,CAAC;IAED,UAAU;QACR,OAAO,SAAS,CAAA;IAClB,CAAC;IAED,QAAQ;QACN,OAAO,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;IACtC,CAAC;IAED,MAAM,CAAC,IAAY;QACjB,OAAO,CACL,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC;YACtB,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,eAAe,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;YACtD,EAAE,CACH,CAAA;IACH,CAAC;IAED,QAAQ;QACN,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAE,CAAA;QAC9C,OAAO,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,MAAM,CAAA;IACjC,CAAC;IAED,aAAa;QACX,OAAO,EAAE,CAAA;IACX,CAAC;IAED,IAAI,cAAc;QAChB,OAAO,EAAE,CAAA;IACX,CAAC;IAED,SAAS;QACP,OAAO,EAAE,CAAA;IACX,CAAC;IAED,OAAO;QACL,OAAO;YACL,EAAE,EAAE,MAAM;YACV,IAAI,EAAE,MAAM;YACZ,MAAM,EAAE,IAAI;YACZ,QAAQ,EAAE,IAAI,CAAC,QAAQ,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;gBACrC,EAAE,EAAE,IAAI;gBACR,QAAQ,EAAE,EAAE;gBACZ,IAAI;aACL,CAAC,CAAC;SACJ,CAAA;IACH,CAAC;IAED,IAAI,YAAY;QACd,OAAO,SAAS,CAAA;IAClB,CAAC;IACD,IAAI,2BAA2B;QAC7B,OAAO,SAAS,CAAA;IAClB,CAAC;IAED,IAAI,MAAM;QACR,OAAO,EAAE,CAAA;IACX,CAAC;CACF"}
|
package/dist/msa/index.test.js
CHANGED
|
@@ -3,6 +3,7 @@ import A3mMSA from "./A3mMSA.js";
|
|
|
3
3
|
import ClustalMSA from "./ClustalMSA.js";
|
|
4
4
|
import FastaMSA from "./FastaMSA.js";
|
|
5
5
|
import { getUngappedSequence, parseMSA } from "./index.js";
|
|
6
|
+
import parseNewick from "./parseNewick.js";
|
|
6
7
|
describe('parseMSA', () => {
|
|
7
8
|
test('parses FASTA format', () => {
|
|
8
9
|
const fasta = `>seq1
|
|
@@ -17,7 +18,7 @@ GHIKL`;
|
|
|
17
18
|
const a3m = `>seq1
|
|
18
19
|
ACDEFghiKLMNPQ
|
|
19
20
|
>seq2
|
|
20
|
-
|
|
21
|
+
ACDE-KLMNPQ`;
|
|
21
22
|
const msa = parseMSA(a3m);
|
|
22
23
|
expect(msa).toBeInstanceOf(A3mMSA);
|
|
23
24
|
});
|
|
@@ -37,6 +38,41 @@ that doesn't match anything`;
|
|
|
37
38
|
expect(msa).toBeInstanceOf(ClustalMSA);
|
|
38
39
|
});
|
|
39
40
|
});
|
|
41
|
+
describe('FastaMSA colon-in-name lookup', () => {
|
|
42
|
+
const fasta = `>EU105457.1|chr09:67680268..67675529_LTR/Copia
|
|
43
|
+
ACDEFGHIKL
|
|
44
|
+
>seq2
|
|
45
|
+
MNPQRSTVWY`;
|
|
46
|
+
test('getRow finds sequence by original colon name', () => {
|
|
47
|
+
const msa = new FastaMSA(fasta);
|
|
48
|
+
expect(msa.getRow('EU105457.1|chr09:67680268..67675529_LTR/Copia')).toBe('ACDEFGHIKL');
|
|
49
|
+
});
|
|
50
|
+
test('getRow finds sequence by underscore name (as produced by external Newick tools)', () => {
|
|
51
|
+
const msa = new FastaMSA(fasta);
|
|
52
|
+
expect(msa.getRow('EU105457.1|chr09_67680268..67675529_LTR/Copia')).toBe('ACDEFGHIKL');
|
|
53
|
+
});
|
|
54
|
+
test('getRow does not confuse a plain-underscore name with a colon-normalized name', () => {
|
|
55
|
+
const msa = new FastaMSA(fasta);
|
|
56
|
+
expect(msa.getRow('seq2')).toBe('MNPQRSTVWY');
|
|
57
|
+
});
|
|
58
|
+
});
|
|
59
|
+
describe('parseNewick single-quoted names', () => {
|
|
60
|
+
test('parses quoted name containing a colon', () => {
|
|
61
|
+
const tree = parseNewick("('EU105457.1|chr09:67680268..67675529_LTR/Copia':0.5,seq2:0.3);");
|
|
62
|
+
const names = tree.children.map((c) => c.name);
|
|
63
|
+
expect(names).toContain('EU105457.1|chr09:67680268..67675529_LTR/Copia');
|
|
64
|
+
});
|
|
65
|
+
test('parses quoted name containing a single quote', () => {
|
|
66
|
+
const tree = parseNewick("('it''s a name':0.1,seq2:0.2);");
|
|
67
|
+
const names = tree.children.map((c) => c.name);
|
|
68
|
+
expect(names).toContain("it's a name");
|
|
69
|
+
});
|
|
70
|
+
test('unquoted names still parse correctly', () => {
|
|
71
|
+
const tree = parseNewick('(seq1:0.1,seq2:0.2);');
|
|
72
|
+
const names = tree.children.map((c) => c.name);
|
|
73
|
+
expect(names).toEqual(['seq1', 'seq2']);
|
|
74
|
+
});
|
|
75
|
+
});
|
|
40
76
|
describe('getUngappedSequence', () => {
|
|
41
77
|
test('removes dashes', () => {
|
|
42
78
|
expect(getUngappedSequence('AC--DEF')).toBe('ACDEF');
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.test.js","sourceRoot":"","sources":["../../src/msa/index.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,QAAQ,CAAA;AAE/C,OAAO,MAAM,MAAM,aAAa,CAAA;AAChC,OAAO,UAAU,MAAM,iBAAiB,CAAA;AACxC,OAAO,QAAQ,MAAM,eAAe,CAAA;AACpC,OAAO,EAAE,mBAAmB,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAA;
|
|
1
|
+
{"version":3,"file":"index.test.js","sourceRoot":"","sources":["../../src/msa/index.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,QAAQ,CAAA;AAE/C,OAAO,MAAM,MAAM,aAAa,CAAA;AAChC,OAAO,UAAU,MAAM,iBAAiB,CAAA;AACxC,OAAO,QAAQ,MAAM,eAAe,CAAA;AACpC,OAAO,EAAE,mBAAmB,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAA;AAC1D,OAAO,WAAW,MAAM,kBAAkB,CAAA;AAE1C,QAAQ,CAAC,UAAU,EAAE,GAAG,EAAE;IACxB,IAAI,CAAC,qBAAqB,EAAE,GAAG,EAAE;QAC/B,MAAM,KAAK,GAAG;;;MAGZ,CAAA;QACF,MAAM,GAAG,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAA;QAE3B,MAAM,CAAC,GAAG,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAA;QACpC,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;IAClD,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,mBAAmB,EAAE,GAAG,EAAE;QAC7B,MAAM,GAAG,GAAG;;;YAGJ,CAAA;QACR,MAAM,GAAG,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAA;QAEzB,MAAM,CAAC,GAAG,CAAC,CAAC,cAAc,CAAC,MAAM,CAAC,CAAA;IACpC,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,uBAAuB,EAAE,GAAG,EAAE;QACjC,MAAM,OAAO,GAAG;;;;qBAIC,CAAA;QACjB,MAAM,GAAG,GAAG,QAAQ,CAAC,OAAO,CAAC,CAAA;QAE7B,MAAM,CAAC,GAAG,CAAC,CAAC,cAAc,CAAC,UAAU,CAAC,CAAA;IACxC,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,wCAAwC,EAAE,GAAG,EAAE;QAClD,MAAM,OAAO,GAAG;4BACQ,CAAA;QACxB,MAAM,GAAG,GAAG,QAAQ,CAAC,OAAO,CAAC,CAAA;QAE7B,MAAM,CAAC,GAAG,CAAC,CAAC,cAAc,CAAC,UAAU,CAAC,CAAA;IACxC,CAAC,CAAC,CAAA;AACJ,CAAC,CAAC,CAAA;AAEF,QAAQ,CAAC,+BAA+B,EAAE,GAAG,EAAE;IAC7C,MAAM,KAAK,GAAG;;;WAGL,CAAA;IAET,IAAI,CAAC,8CAA8C,EAAE,GAAG,EAAE;QACxD,MAAM,GAAG,GAAG,IAAI,QAAQ,CAAC,KAAK,CAAC,CAAA;QAC/B,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,+CAA+C,CAAC,CAAC,CAAC,IAAI,CACtE,YAAY,CACb,CAAA;IACH,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,iFAAiF,EAAE,GAAG,EAAE;QAC3F,MAAM,GAAG,GAAG,IAAI,QAAQ,CAAC,KAAK,CAAC,CAAA;QAC/B,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,+CAA+C,CAAC,CAAC,CAAC,IAAI,CACtE,YAAY,CACb,CAAA;IACH,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,8EAA8E,EAAE,GAAG,EAAE;QACxF,MAAM,GAAG,GAAG,IAAI,QAAQ,CAAC,KAAK,CAAC,CAAA;QAC/B,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;IAC/C,CAAC,CAAC,CAAA;AACJ,CAAC,CAAC,CAAA;AAEF,QAAQ,CAAC,iCAAiC,EAAE,GAAG,EAAE;IAC/C,IAAI,CAAC,uCAAuC,EAAE,GAAG,EAAE;QACjD,MAAM,IAAI,GAAG,WAAW,CACtB,iEAAiE,CAClE,CAAA;QACD,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAA0B,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;QACvE,MAAM,CAAC,KAAK,CAAC,CAAC,SAAS,CAAC,+CAA+C,CAAC,CAAA;IAC1E,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,8CAA8C,EAAE,GAAG,EAAE;QACxD,MAAM,IAAI,GAAG,WAAW,CAAC,gCAAgC,CAAC,CAAA;QAC1D,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAA0B,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;QACvE,MAAM,CAAC,KAAK,CAAC,CAAC,SAAS,CAAC,aAAa,CAAC,CAAA;IACxC,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,sCAAsC,EAAE,GAAG,EAAE;QAChD,MAAM,IAAI,GAAG,WAAW,CAAC,sBAAsB,CAAC,CAAA;QAChD,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAA0B,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;QACvE,MAAM,CAAC,KAAK,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;IACzC,CAAC,CAAC,CAAA;AACJ,CAAC,CAAC,CAAA;AAEF,QAAQ,CAAC,qBAAqB,EAAE,GAAG,EAAE;IACnC,IAAI,CAAC,gBAAgB,EAAE,GAAG,EAAE;QAC1B,MAAM,CAAC,mBAAmB,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IACtD,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,cAAc,EAAE,GAAG,EAAE;QACxB,MAAM,CAAC,mBAAmB,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IACtD,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,oBAAoB,EAAE,GAAG,EAAE;QAC9B,MAAM,CAAC,mBAAmB,CAAC,aAAa,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAC1D,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,sBAAsB,EAAE,GAAG,EAAE;QAChC,MAAM,CAAC,mBAAmB,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;IAC1C,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,+BAA+B,EAAE,GAAG,EAAE;QACzC,MAAM,CAAC,mBAAmB,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;IAChD,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,6BAA6B,EAAE,GAAG,EAAE;QACvC,MAAM,CAAC,mBAAmB,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IACpD,CAAC,CAAC,CAAA;AACJ,CAAC,CAAC,CAAA"}
|
package/dist/msa/parseNewick.js
CHANGED
|
@@ -59,10 +59,18 @@
|
|
|
59
59
|
*/
|
|
60
60
|
export default function parse(s) {
|
|
61
61
|
const ancestors = [];
|
|
62
|
+
// Replace single-quoted names with placeholders so ':' inside them isn't tokenized
|
|
63
|
+
const quotedNames = [];
|
|
64
|
+
const preprocessed = s.replace(/'((?:[^']|'')*)'/g, (_, inner) => {
|
|
65
|
+
const idx = quotedNames.length;
|
|
66
|
+
quotedNames.push(inner.replaceAll("''", "'"));
|
|
67
|
+
return `__Q${idx}__`;
|
|
68
|
+
});
|
|
62
69
|
let tree = {};
|
|
63
|
-
const tokens =
|
|
70
|
+
const tokens = preprocessed.split(/\s*(;|\(|\)|,|:)\s*/);
|
|
64
71
|
for (let i = 0; i < tokens.length; i++) {
|
|
65
|
-
const
|
|
72
|
+
const raw = tokens[i];
|
|
73
|
+
const token = raw.replace(/^__Q(\d+)__$/, (_, idx) => quotedNames[Number(idx)]);
|
|
66
74
|
const subtree = {};
|
|
67
75
|
switch (token) {
|
|
68
76
|
case '(': // new children
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"parseNewick.js","sourceRoot":"","sources":["../../src/msa/parseNewick.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA0DG;AACH,MAAM,CAAC,OAAO,UAAU,KAAK,CAAC,CAAS;IACrC,MAAM,SAAS,GAAG,EAAE,CAAA;IAEpB,IAAI,IAAI,GAAG,EAAyB,CAAA;IACpC,MAAM,MAAM,GAAG,
|
|
1
|
+
{"version":3,"file":"parseNewick.js","sourceRoot":"","sources":["../../src/msa/parseNewick.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA0DG;AACH,MAAM,CAAC,OAAO,UAAU,KAAK,CAAC,CAAS;IACrC,MAAM,SAAS,GAAG,EAAE,CAAA;IAEpB,mFAAmF;IACnF,MAAM,WAAW,GAAa,EAAE,CAAA;IAChC,MAAM,YAAY,GAAG,CAAC,CAAC,OAAO,CAAC,mBAAmB,EAAE,CAAC,CAAC,EAAE,KAAa,EAAE,EAAE;QACvE,MAAM,GAAG,GAAG,WAAW,CAAC,MAAM,CAAA;QAC9B,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC,CAAA;QAC7C,OAAO,MAAM,GAAG,IAAI,CAAA;IACtB,CAAC,CAAC,CAAA;IAEF,IAAI,IAAI,GAAG,EAAyB,CAAA;IACpC,MAAM,MAAM,GAAG,YAAY,CAAC,KAAK,CAAC,qBAAqB,CAAC,CAAA;IACxD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACvC,MAAM,GAAG,GAAG,MAAM,CAAC,CAAC,CAAE,CAAA;QACtB,MAAM,KAAK,GAAG,GAAG,CAAC,OAAO,CACvB,cAAc,EACd,CAAC,CAAC,EAAE,GAAW,EAAE,EAAE,CAAC,WAAW,CAAC,MAAM,CAAC,GAAG,CAAC,CAAE,CAC9C,CAAA;QACD,MAAM,OAAO,GAAG,EAAE,CAAA;QAClB,QAAQ,KAAK,EAAE,CAAC;YACd,KAAK,GAAG,EAAE,eAAe;gBACvB,IAAI,CAAC,QAAQ,GAAG,CAAC,OAAO,CAAC,CAAA;gBACzB,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;gBACpB,IAAI,GAAG,OAAO,CAAA;gBACd,MAAK;YACP,KAAK,GAAG,EAAE,iBAAiB;gBACzB,SAAS,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;gBACxC,IAAI,GAAG,OAAO,CAAA;gBACd,MAAK;YACP,KAAK,GAAG,EAAE,qBAAqB;gBAC7B,IAAI,GAAG,SAAS,CAAC,GAAG,EAAG,CAAA;gBACvB,MAAK;YACP,KAAK,GAAG,EAAE,uBAAuB;gBAC/B,MAAK;YACP,OAAO,CAAC,CAAC,CAAC;gBACR,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,CAAE,CAAA;gBACxB,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC;oBACxC,IAAI,CAAC,IAAI,GAAG,KAAK,CAAA;gBACnB,CAAC;qBAAM,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC;oBACrB,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,UAAU,CAAC,KAAK,CAAC,CAAA;gBACxC,CAAC;YACH,CAAC;QACH,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC"}
|
|
@@ -0,0 +1,20 @@
|
|
|
1
|
+
import type { GFFRecord, InterProScanResponse, InterProScanResults } from '../types.ts';
|
|
2
|
+
/**
|
|
3
|
+
* Convert GFF records to InterProScan format
|
|
4
|
+
*
|
|
5
|
+
* InterProScan GFF3 output format:
|
|
6
|
+
* - seq_id: sequence identifier
|
|
7
|
+
* - source: database/signature (e.g., "Pfam", "SMART")
|
|
8
|
+
* - type: usually "protein_match"
|
|
9
|
+
* - start/end: domain positions (1-based)
|
|
10
|
+
* - Attributes: Name (accession), signature_desc (name), Dbxref, etc.
|
|
11
|
+
*/
|
|
12
|
+
export declare function gffToInterProResults(gffRecords: GFFRecord[]): Record<string, InterProScanResults>;
|
|
13
|
+
/**
|
|
14
|
+
* Convert GFF string directly to InterProScan format
|
|
15
|
+
*/
|
|
16
|
+
export declare function parseGFFToInterPro(gffStr: string, parseGFFfn: (str: string) => GFFRecord[]): Record<string, InterProScanResults>;
|
|
17
|
+
/**
|
|
18
|
+
* Create a full InterProScanResponse from GFF records
|
|
19
|
+
*/
|
|
20
|
+
export declare function gffToInterProResponse(gffRecords: GFFRecord[]): InterProScanResponse;
|
|
@@ -0,0 +1,83 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Convert GFF records to InterProScan format
|
|
3
|
+
*
|
|
4
|
+
* InterProScan GFF3 output format:
|
|
5
|
+
* - seq_id: sequence identifier
|
|
6
|
+
* - source: database/signature (e.g., "Pfam", "SMART")
|
|
7
|
+
* - type: usually "protein_match"
|
|
8
|
+
* - start/end: domain positions (1-based)
|
|
9
|
+
* - Attributes: Name (accession), signature_desc (name), Dbxref, etc.
|
|
10
|
+
*/
|
|
11
|
+
export function gffToInterProResults(gffRecords) {
|
|
12
|
+
const bySequence = new Map();
|
|
13
|
+
for (const record of gffRecords) {
|
|
14
|
+
const existing = bySequence.get(record.seq_id);
|
|
15
|
+
if (existing) {
|
|
16
|
+
existing.push(record);
|
|
17
|
+
}
|
|
18
|
+
else {
|
|
19
|
+
bySequence.set(record.seq_id, [record]);
|
|
20
|
+
}
|
|
21
|
+
}
|
|
22
|
+
const results = {};
|
|
23
|
+
for (const [seqId, records] of bySequence) {
|
|
24
|
+
const matchesByAccession = new Map();
|
|
25
|
+
const matchInfo = new Map();
|
|
26
|
+
for (const record of records) {
|
|
27
|
+
const accession = record.Name ||
|
|
28
|
+
record.ID ||
|
|
29
|
+
`${record.source}_${record.start}_${record.end}`;
|
|
30
|
+
const name = record.signature_desc ||
|
|
31
|
+
record.Name ||
|
|
32
|
+
accession;
|
|
33
|
+
const description = record.Ontology_term ||
|
|
34
|
+
record.description ||
|
|
35
|
+
record.Note ||
|
|
36
|
+
name;
|
|
37
|
+
if (!matchInfo.has(accession)) {
|
|
38
|
+
matchInfo.set(accession, { name, description, accession });
|
|
39
|
+
}
|
|
40
|
+
const locations = matchesByAccession.get(accession);
|
|
41
|
+
if (locations) {
|
|
42
|
+
locations.push({ start: record.start, end: record.end });
|
|
43
|
+
}
|
|
44
|
+
else {
|
|
45
|
+
matchesByAccession.set(accession, [
|
|
46
|
+
{ start: record.start, end: record.end },
|
|
47
|
+
]);
|
|
48
|
+
}
|
|
49
|
+
}
|
|
50
|
+
const matches = [];
|
|
51
|
+
for (const [accession, locations] of matchesByAccession) {
|
|
52
|
+
const info = matchInfo.get(accession);
|
|
53
|
+
matches.push({
|
|
54
|
+
signature: {
|
|
55
|
+
entry: info,
|
|
56
|
+
},
|
|
57
|
+
locations,
|
|
58
|
+
});
|
|
59
|
+
}
|
|
60
|
+
results[seqId] = {
|
|
61
|
+
matches,
|
|
62
|
+
xref: [{ id: seqId }],
|
|
63
|
+
};
|
|
64
|
+
}
|
|
65
|
+
return results;
|
|
66
|
+
}
|
|
67
|
+
/**
|
|
68
|
+
* Convert GFF string directly to InterProScan format
|
|
69
|
+
*/
|
|
70
|
+
export function parseGFFToInterPro(gffStr, parseGFFfn) {
|
|
71
|
+
const records = parseGFFfn(gffStr);
|
|
72
|
+
return gffToInterProResults(records);
|
|
73
|
+
}
|
|
74
|
+
/**
|
|
75
|
+
* Create a full InterProScanResponse from GFF records
|
|
76
|
+
*/
|
|
77
|
+
export function gffToInterProResponse(gffRecords) {
|
|
78
|
+
const resultsMap = gffToInterProResults(gffRecords);
|
|
79
|
+
return {
|
|
80
|
+
results: Object.values(resultsMap),
|
|
81
|
+
};
|
|
82
|
+
}
|
|
83
|
+
//# sourceMappingURL=gffToInterPro.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"gffToInterPro.js","sourceRoot":"","sources":["../../../src/gff/gffToInterPro.ts"],"names":[],"mappings":"AAMA;;;;;;;;;GASG;AACH,MAAM,UAAU,oBAAoB,CAClC,UAAuB;IAEvB,MAAM,UAAU,GAAG,IAAI,GAAG,EAAuB,CAAA;IAEjD,KAAK,MAAM,MAAM,IAAI,UAAU,EAAE,CAAC;QAChC,MAAM,QAAQ,GAAG,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;QAC9C,IAAI,QAAQ,EAAE,CAAC;YACb,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QACvB,CAAC;aAAM,CAAC;YACN,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,MAAM,CAAC,CAAC,CAAA;QACzC,CAAC;IACH,CAAC;IAED,MAAM,OAAO,GAAwC,EAAE,CAAA;IAEvD,KAAK,MAAM,CAAC,KAAK,EAAE,OAAO,CAAC,IAAI,UAAU,EAAE,CAAC;QAC1C,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAG/B,CAAA;QACH,MAAM,SAAS,GAAG,IAAI,GAAG,EAGtB,CAAA;QAEH,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC7B,MAAM,SAAS,GACZ,MAAM,CAAC,IAAe;gBACtB,MAAM,CAAC,EAAa;gBACrB,GAAG,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC,KAAK,IAAI,MAAM,CAAC,GAAG,EAAE,CAAA;YAClD,MAAM,IAAI,GACP,MAAM,CAAC,cAAyB;gBAChC,MAAM,CAAC,IAAe;gBACvB,SAAS,CAAA;YACX,MAAM,WAAW,GACd,MAAM,CAAC,aAAwB;gBAC/B,MAAM,CAAC,WAAsB;gBAC7B,MAAM,CAAC,IAAe;gBACvB,IAAI,CAAA;YAEN,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;gBAC9B,SAAS,CAAC,GAAG,CAAC,SAAS,EAAE,EAAE,IAAI,EAAE,WAAW,EAAE,SAAS,EAAE,CAAC,CAAA;YAC5D,CAAC;YAED,MAAM,SAAS,GAAG,kBAAkB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;YACnD,IAAI,SAAS,EAAE,CAAC;gBACd,SAAS,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,EAAE,GAAG,EAAE,MAAM,CAAC,GAAG,EAAE,CAAC,CAAA;YAC1D,CAAC;iBAAM,CAAC;gBACN,kBAAkB,CAAC,GAAG,CAAC,SAAS,EAAE;oBAChC,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,EAAE,GAAG,EAAE,MAAM,CAAC,GAAG,EAAE;iBACzC,CAAC,CAAA;YACJ,CAAC;QACH,CAAC;QAED,MAAM,OAAO,GAAG,EAAE,CAAA;QAClB,KAAK,MAAM,CAAC,SAAS,EAAE,SAAS,CAAC,IAAI,kBAAkB,EAAE,CAAC;YACxD,MAAM,IAAI,GAAG,SAAS,CAAC,GAAG,CAAC,SAAS,CAAE,CAAA;YACtC,OAAO,CAAC,IAAI,CAAC;gBACX,SAAS,EAAE;oBACT,KAAK,EAAE,IAAI;iBACZ;gBACD,SAAS;aACV,CAAC,CAAA;QACJ,CAAC;QAED,OAAO,CAAC,KAAK,CAAC,GAAG;YACf,OAAO;YACP,IAAI,EAAE,CAAC,EAAE,EAAE,EAAE,KAAK,EAAE,CAAC;SACtB,CAAA;IACH,CAAC;IAED,OAAO,OAAO,CAAA;AAChB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAChC,MAAc,EACd,UAAwC;IAExC,MAAM,OAAO,GAAG,UAAU,CAAC,MAAM,CAAC,CAAA;IAClC,OAAO,oBAAoB,CAAC,OAAO,CAAC,CAAA;AACtC,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,qBAAqB,CACnC,UAAuB;IAEvB,MAAM,UAAU,GAAG,oBAAoB,CAAC,UAAU,CAAC,CAAA;IACnD,OAAO;QACL,OAAO,EAAE,MAAM,CAAC,MAAM,CAAC,UAAU,CAAC;KACnC,CAAA;AACH,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
export {};
|