msa-parsers 5.3.0 → 5.4.0

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/index.d.ts CHANGED
@@ -1,5 +1,5 @@
1
1
  export * from './types.ts';
2
2
  export { generateNodeIds } from './util.ts';
3
3
  export { A3mMSA, ClustalMSA, EmfMSA, FastaMSA, StockholmMSA, getUngappedSequence, parseEmfTree, parseMSA, parseNewick, stockholmSniff, } from './msa/index.ts';
4
- export type { MSAParserType } from './msa/index.ts';
4
+ export type { MSAFormat, MSAParserType } from './msa/index.ts';
5
5
  export { gffToInterProResponse, gffToInterProResults, interProResponseToGFF, interProToGFF, parseGFF, } from './gff/index.ts';
@@ -76,6 +76,14 @@ export default class A3mMSA extends BaseMSA {
76
76
  if (seqs.length < 2) {
77
77
  return false;
78
78
  }
79
+ // A true aligned FASTA (e.g. impg's fasta-aln output from soft-masked
80
+ // references) has all rows the same total length. A3M's defining feature is
81
+ // variable-length rows, since lowercase inserts live outside the fixed match
82
+ // grid. So equal-length rows are an alignment, not A3M, even when lowercase
83
+ // soft-masking is present.
84
+ if (new Set(seqs.map(s => s.length)).size === 1) {
85
+ return false;
86
+ }
79
87
  let hasLowercase = false;
80
88
  let firstMatchLen = -1;
81
89
  let sameMatchLength = true;
@@ -1 +1 @@
1
- {"version":3,"file":"A3mMSA.js","sourceRoot":"","sources":["../../src/msa/A3mMSA.ts"],"names":[],"mappings":"AAAA,OAAO,OAAO,MAAM,cAAc,CAAA;AAElC;;;;;;;;;;;;;;;;;;;;GAoBG;AAEH,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,MAAO,SAAQ,OAAO;IACjC,GAAG,CAAqC;IACxC,YAAY,CAAU;IAE9B,YAAY,IAAY;QACtB,KAAK,EAAE,CAAA;QACP,MAAM,OAAO,GAAa,EAAE,CAAA;QAC5B,MAAM,KAAK,GAAa,EAAE,CAAA;QAE1B,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,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAA;YAC7D,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,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,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,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;IAEO,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,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,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAE,CAAC,CAAA;oBACrB,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,CAAC,EAAE,CAAA;gBACL,CAAC;qBAAM,IAAI,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;oBACzB,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,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,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,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,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,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,MAAM,CAAC,IAAI,CAAC,SAAS,IAAI,GAAG,CAAC,CAAA;oBAE7B,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,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;oBAEtB,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,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,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;CACF"}
1
+ {"version":3,"file":"A3mMSA.js","sourceRoot":"","sources":["../../src/msa/A3mMSA.ts"],"names":[],"mappings":"AAAA,OAAO,OAAO,MAAM,cAAc,CAAA;AAElC;;;;;;;;;;;;;;;;;;;;GAoBG;AAEH,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,MAAO,SAAQ,OAAO;IACjC,GAAG,CAAqC;IACxC,YAAY,CAAU;IAE9B,YAAY,IAAY;QACtB,KAAK,EAAE,CAAA;QACP,MAAM,OAAO,GAAa,EAAE,CAAA;QAC5B,MAAM,KAAK,GAAa,EAAE,CAAA;QAE1B,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,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAA;YAC7D,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,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,sEAAsE;QACtE,4EAA4E;QAC5E,6EAA6E;QAC7E,4EAA4E;QAC5E,2BAA2B;QAC3B,IAAI,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;YAChD,OAAO,KAAK,CAAA;QACd,CAAC;QAED,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,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;IAEO,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,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,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAE,CAAC,CAAA;oBACrB,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,CAAC,EAAE,CAAA;gBACL,CAAC;qBAAM,IAAI,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;oBACzB,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,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,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,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,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,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,MAAM,CAAC,IAAI,CAAC,SAAS,IAAI,GAAG,CAAC,CAAA;oBAE7B,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,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;oBAEtB,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,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,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;CACF"}
@@ -85,9 +85,19 @@ export default class StockholmMSA extends BaseMSA {
85
85
  id: 'secondaryStruct',
86
86
  name: 'Secondary-structure',
87
87
  data: this.secondaryStructureConsensus,
88
+ // WUSS notation pairs base-paired columns as open/close brackets of
89
+ // several nesting types (<>, (), [], {}); color all opens one way and
90
+ // closes the other so every helix (e.g. the tRNA acceptor stem, which
91
+ // is written with parens) is highlighted, not just the <> arms.
88
92
  customColorScheme: {
89
- '>': 'pink',
90
93
  '<': 'lightblue',
94
+ '(': 'lightblue',
95
+ '[': 'lightblue',
96
+ '{': 'lightblue',
97
+ '>': 'pink',
98
+ ')': 'pink',
99
+ ']': 'pink',
100
+ '}': 'pink',
91
101
  },
92
102
  },
93
103
  ];
@@ -1 +1 @@
1
- {"version":3,"file":"StockholmMSA.js","sourceRoot":"","sources":["../../src/msa/StockholmMSA.ts"],"names":[],"mappings":"AAAA,OAAO,WAAW,MAAM,kBAAkB,CAAA;AAC1C,OAAO,EAAE,QAAQ,EAAE,MAAM,sBAAsB,CAAA;AAC/C,OAAO,EAAE,eAAe,EAAE,MAAM,YAAY,CAAA;AAC5C,OAAO,OAAO,MAAM,cAAc,CAAA;AAKlC,MAAM,CAAC,OAAO,OAAO,YAAa,SAAQ,OAAO;IACvC,IAAI,CAAiB;IACrB,GAAG,CAAe;IAE1B,YAAY,IAAY,EAAE,gBAAwB;QAChD,KAAK,EAAE,CAAA;QACP,MAAM,GAAG,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAA;QAC1B,IAAI,CAAC,IAAI,GAAG,GAAG,CAAA;QACf,IAAI,CAAC,GAAG,GAAG,GAAG,CAAC,gBAAgB,CAAE,CAAA;IACnC,CAAC;IAED,MAAM;QACJ,OAAO,IAAI,CAAC,GAAG,CAAA;IACjB,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,CAAE,CAAA;QAC9C,OAAO,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,MAAM,CAAA;IACjC,CAAC;IAED,IAAI,cAAc;QAChB,OAAO,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,aAAa,GAAG,GAAG,CAAC,EAAE,CAAC,CAAA;IAC9E,CAAC;IAED,SAAS;QACP,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAA;QACtC,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAA;QACtC,OAAO;YACL,OAAO,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE;YACpB,UAAU,EAAE,MAAM,CAAC,WAAW,CAC5B,MAAM,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CACrD;YACD,MAAM,EAAE,MAAM,CAAC,WAAW,CACxB,MAAM,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAC7D;SACF,CAAA;IACH,CAAC;IAED,UAAU,CAAC,OAAe;QACxB,OAAO;YACL,IAAI,EAAE,OAAO;YACb,SAAS,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC;YACzC,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC;SAC9C,CAAA;IACH,CAAC;IAED,QAAQ;QACN,OAAO,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;IACtC,CAAC;IAED,aAAa;QACX,MAAM,QAAQ,GAAG,sCAAsC,CAAA;QACvD,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAA;QACtC,MAAM,MAAM,GAGR,EAAE,CAAA;QACN,KAAK,MAAM,CAAC,EAAE,EAAE,MAAM,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,SAAS,CAAC,EAAE,CAAC;YACrD,KAAK,MAAM,EAAE,IAAI,MAAM,EAAE,CAAC;gBACxB,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;gBAC/B,IAAI,KAAK,EAAE,CAAC;oBACV,CAAC;oBAAA,CAAC,MAAM,CAAC,EAAE,CAAC,KAAK,EAAE,CAAC,CAAC,IAAI,CAAC;wBACxB,GAAG,EAAE,KAAK,CAAC,CAAC,CAAE,CAAC,WAAW,EAAE;wBAC5B,KAAK,EAAE,KAAK,CAAC,CAAC,CAAE;wBAChB,QAAQ,EAAE,CAAC,KAAK,CAAC,CAAC,CAAE;wBACpB,MAAM,EAAE,CAAC,KAAK,CAAC,CAAC,CAAE;qBACnB,CAAC,CAAA;gBACJ,CAAC;YACH,CAAC;QACH,CAAC;QACD,OAAO,MAAM,CAAA;IACf,CAAC;IAED,OAAO;QACL,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAA;QAChC,OAAO,IAAI,CAAC,CAAC,CAAC,eAAe,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,EAAE,CAAA;IACpE,CAAC;IAED,IAAI,YAAY;QACd,OAAO,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,QAAQ,CAAA;IAC7B,CAAC;IAED,IAAI,2BAA2B;QAC7B,OAAO,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,CAAA;IAC5B,CAAC;IAED,IAAI,MAAM;QAMR,OAAO;YACL;gBACE,EAAE,EAAE,cAAc;gBAClB,IAAI,EAAE,oBAAoB;gBAC1B,IAAI,EAAE,IAAI,CAAC,YAAY;gBACvB,iBAAiB,EAAE,EAAE;aACtB;YACD;gBACE,EAAE,EAAE,iBAAiB;gBACrB,IAAI,EAAE,qBAAqB;gBAC3B,IAAI,EAAE,IAAI,CAAC,2BAA2B;gBACtC,iBAAiB,EAAE;oBACjB,GAAG,EAAE,MAAM;oBACX,GAAG,EAAE,WAAW;iBACjB;aACF;SACF,CAAA;IACH,CAAC;CACF;AAED,OAAO,EAAE,KAAK,IAAI,cAAc,EAAE,MAAM,sBAAsB,CAAA"}
1
+ {"version":3,"file":"StockholmMSA.js","sourceRoot":"","sources":["../../src/msa/StockholmMSA.ts"],"names":[],"mappings":"AAAA,OAAO,WAAW,MAAM,kBAAkB,CAAA;AAC1C,OAAO,EAAE,QAAQ,EAAE,MAAM,sBAAsB,CAAA;AAC/C,OAAO,EAAE,eAAe,EAAE,MAAM,YAAY,CAAA;AAC5C,OAAO,OAAO,MAAM,cAAc,CAAA;AAKlC,MAAM,CAAC,OAAO,OAAO,YAAa,SAAQ,OAAO;IACvC,IAAI,CAAiB;IACrB,GAAG,CAAe;IAE1B,YAAY,IAAY,EAAE,gBAAwB;QAChD,KAAK,EAAE,CAAA;QACP,MAAM,GAAG,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAA;QAC1B,IAAI,CAAC,IAAI,GAAG,GAAG,CAAA;QACf,IAAI,CAAC,GAAG,GAAG,GAAG,CAAC,gBAAgB,CAAE,CAAA;IACnC,CAAC;IAED,MAAM;QACJ,OAAO,IAAI,CAAC,GAAG,CAAA;IACjB,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,CAAE,CAAA;QAC9C,OAAO,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,MAAM,CAAA;IACjC,CAAC;IAED,IAAI,cAAc;QAChB,OAAO,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,aAAa,GAAG,GAAG,CAAC,EAAE,CAAC,CAAA;IAC9E,CAAC;IAED,SAAS;QACP,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAA;QACtC,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAA;QACtC,OAAO;YACL,OAAO,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE;YACpB,UAAU,EAAE,MAAM,CAAC,WAAW,CAC5B,MAAM,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CACrD;YACD,MAAM,EAAE,MAAM,CAAC,WAAW,CACxB,MAAM,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAC7D;SACF,CAAA;IACH,CAAC;IAED,UAAU,CAAC,OAAe;QACxB,OAAO;YACL,IAAI,EAAE,OAAO;YACb,SAAS,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC;YACzC,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC;SAC9C,CAAA;IACH,CAAC;IAED,QAAQ;QACN,OAAO,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;IACtC,CAAC;IAED,aAAa;QACX,MAAM,QAAQ,GAAG,sCAAsC,CAAA;QACvD,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAA;QACtC,MAAM,MAAM,GAGR,EAAE,CAAA;QACN,KAAK,MAAM,CAAC,EAAE,EAAE,MAAM,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,SAAS,CAAC,EAAE,CAAC;YACrD,KAAK,MAAM,EAAE,IAAI,MAAM,EAAE,CAAC;gBACxB,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;gBAC/B,IAAI,KAAK,EAAE,CAAC;oBACV,CAAC;oBAAA,CAAC,MAAM,CAAC,EAAE,CAAC,KAAK,EAAE,CAAC,CAAC,IAAI,CAAC;wBACxB,GAAG,EAAE,KAAK,CAAC,CAAC,CAAE,CAAC,WAAW,EAAE;wBAC5B,KAAK,EAAE,KAAK,CAAC,CAAC,CAAE;wBAChB,QAAQ,EAAE,CAAC,KAAK,CAAC,CAAC,CAAE;wBACpB,MAAM,EAAE,CAAC,KAAK,CAAC,CAAC,CAAE;qBACnB,CAAC,CAAA;gBACJ,CAAC;YACH,CAAC;QACH,CAAC;QACD,OAAO,MAAM,CAAA;IACf,CAAC;IAED,OAAO;QACL,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAA;QAChC,OAAO,IAAI,CAAC,CAAC,CAAC,eAAe,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,EAAE,CAAA;IACpE,CAAC;IAED,IAAI,YAAY;QACd,OAAO,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,QAAQ,CAAA;IAC7B,CAAC;IAED,IAAI,2BAA2B;QAC7B,OAAO,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,CAAA;IAC5B,CAAC;IAED,IAAI,MAAM;QAMR,OAAO;YACL;gBACE,EAAE,EAAE,cAAc;gBAClB,IAAI,EAAE,oBAAoB;gBAC1B,IAAI,EAAE,IAAI,CAAC,YAAY;gBACvB,iBAAiB,EAAE,EAAE;aACtB;YACD;gBACE,EAAE,EAAE,iBAAiB;gBACrB,IAAI,EAAE,qBAAqB;gBAC3B,IAAI,EAAE,IAAI,CAAC,2BAA2B;gBACtC,oEAAoE;gBACpE,sEAAsE;gBACtE,sEAAsE;gBACtE,gEAAgE;gBAChE,iBAAiB,EAAE;oBACjB,GAAG,EAAE,WAAW;oBAChB,GAAG,EAAE,WAAW;oBAChB,GAAG,EAAE,WAAW;oBAChB,GAAG,EAAE,WAAW;oBAChB,GAAG,EAAE,MAAM;oBACX,GAAG,EAAE,MAAM;oBACX,GAAG,EAAE,MAAM;oBACX,GAAG,EAAE,MAAM;iBACZ;aACF;SACF,CAAA;IACH,CAAC;CACF;AAED,OAAO,EAAE,KAAK,IAAI,cAAc,EAAE,MAAM,sBAAsB,CAAA"}
@@ -0,0 +1 @@
1
+ export {};
@@ -0,0 +1,49 @@
1
+ import { describe, expect, test } from 'vitest';
2
+ import A3mMSA from "./A3mMSA.js";
3
+ import FastaMSA from "./FastaMSA.js";
4
+ import { parseMSA } from "./index.js";
5
+ // impg (https://github.com/pangenome/impg) pipes `impg query -o fasta-aln` into
6
+ // scripts/faln2html.py, which feeds the FASTA string to
7
+ // MSAModelF().create({ data: { msa } }). These tests pin the behavior that
8
+ // integration relies on.
9
+ describe('impg fasta-aln integration', () => {
10
+ // PanSN-style names + region coordinates with colons, dash gaps, uppercase
11
+ const block = `>HG002#1#chr1:1000-1040
12
+ ACGTACGT--ACGTNNNNACGTACGTACGTACGTACGT
13
+ >HG003#1#chr1:1000-1040
14
+ ACGTAC-TGGACGTNNNNACGTAC--ACGTACGTACGT
15
+ >CHM13#0#chr1:1000-1040
16
+ ACGTACGTGGACGTNNNNACGT----ACGTACGTACGT`;
17
+ test('uppercase block is parsed as FASTA, not A3M/Clustal', () => {
18
+ const msa = parseMSA(block);
19
+ expect(msa).toBeInstanceOf(FastaMSA);
20
+ expect(msa.getNames()).toEqual([
21
+ 'HG002#1#chr1:1000-1040',
22
+ 'HG003#1#chr1:1000-1040',
23
+ 'CHM13#0#chr1:1000-1040',
24
+ ]);
25
+ expect(msa.getWidth()).toBe(38);
26
+ });
27
+ test('rows are retrievable by their colon-containing region name', () => {
28
+ const msa = parseMSA(block);
29
+ expect(msa.getRow('HG002#1#chr1:1000-1040')).toBe('ACGTACGT--ACGTNNNNACGTACGTACGTACGTACGT');
30
+ });
31
+ // If impg extracts from a soft-masked reference FASTA, aligned rows carry
32
+ // lowercase bases. A3M sniff runs before FASTA and triggers on
33
+ // hasLowercase && equal upper+dash match-length across rows -- which these
34
+ // rows satisfy by coincidence. The rows are all the same total length though,
35
+ // so they are a real alignment and must parse as FASTA, not A3M (A3M would
36
+ // strip the lowercase into separate insert columns and break alignment).
37
+ const softMasked = `>HG002#1#chr1:1000-1030
38
+ acgtACGTAC--GTACGTACGTACGTacgt
39
+ >HG003#1#chr1:1000-1030
40
+ ACGTacgtAC--GTACGTacgtACGTACGT`;
41
+ test('soft-masked equal-length block parses as FASTA, not A3M', () => {
42
+ const msa = parseMSA(softMasked);
43
+ expect(msa).toBeInstanceOf(FastaMSA);
44
+ expect(msa).not.toBeInstanceOf(A3mMSA);
45
+ const rows = msa.getNames().map(n => msa.getRow(n));
46
+ expect(new Set(rows.map(r => r.length)).size).toBe(1);
47
+ });
48
+ });
49
+ //# sourceMappingURL=impgIntegration.test.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"impgIntegration.test.js","sourceRoot":"","sources":["../../src/msa/impgIntegration.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,QAAQ,MAAM,eAAe,CAAA;AACpC,OAAO,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAA;AAErC,gFAAgF;AAChF,wDAAwD;AACxD,2EAA2E;AAC3E,yBAAyB;AAEzB,QAAQ,CAAC,4BAA4B,EAAE,GAAG,EAAE;IAC1C,2EAA2E;IAC3E,MAAM,KAAK,GAAG;;;;;uCAKuB,CAAA;IAErC,IAAI,CAAC,qDAAqD,EAAE,GAAG,EAAE;QAC/D,MAAM,GAAG,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAA;QAC3B,MAAM,CAAC,GAAG,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAA;QACpC,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC;YAC7B,wBAAwB;YACxB,wBAAwB;YACxB,wBAAwB;SACzB,CAAC,CAAA;QACF,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;IACjC,CAAC,CAAC,CAAA;IAEF,IAAI,CAAC,4DAA4D,EAAE,GAAG,EAAE;QACtE,MAAM,GAAG,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAA;QAC3B,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,wBAAwB,CAAC,CAAC,CAAC,IAAI,CAC/C,wCAAwC,CACzC,CAAA;IACH,CAAC,CAAC,CAAA;IAEF,0EAA0E;IAC1E,+DAA+D;IAC/D,2EAA2E;IAC3E,8EAA8E;IAC9E,2EAA2E;IAC3E,yEAAyE;IACzE,MAAM,UAAU,GAAG;;;+BAGU,CAAA;IAE7B,IAAI,CAAC,yDAAyD,EAAE,GAAG,EAAE;QACnE,MAAM,GAAG,GAAG,QAAQ,CAAC,UAAU,CAAC,CAAA;QAChC,MAAM,CAAC,GAAG,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAA;QACpC,MAAM,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,cAAc,CAAC,MAAM,CAAC,CAAA;QACtC,MAAM,IAAI,GAAG,GAAG,CAAC,QAAQ,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAA;QACnD,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;IACvD,CAAC,CAAC,CAAA;AACJ,CAAC,CAAC,CAAA"}
@@ -6,7 +6,8 @@ import StockholmMSA from './StockholmMSA.ts';
6
6
  export { parseEmfTree } from 'emf-js';
7
7
  export { default as parseNewick } from './parseNewick.ts';
8
8
  export type MSAParserType = StockholmMSA | A3mMSA | FastaMSA | EmfMSA | ClustalMSA;
9
- export declare function parseMSA(text: string, currentAlignment?: number): MSAParserType;
9
+ export type MSAFormat = 'stockholm' | 'a3m' | 'fasta' | 'emf' | 'clustal';
10
+ export declare function parseMSA(text: string, currentAlignment?: number, format?: MSAFormat): MSAParserType;
10
11
  /**
11
12
  * Get ungapped sequence from an aligned row
12
13
  */
package/dist/msa/index.js CHANGED
@@ -5,7 +5,26 @@ import FastaMSA from "./FastaMSA.js";
5
5
  import StockholmMSA, { stockholmSniff } from "./StockholmMSA.js";
6
6
  export { parseEmfTree } from 'emf-js';
7
7
  export { default as parseNewick } from "./parseNewick.js";
8
- export function parseMSA(text, currentAlignment = 0) {
8
+ // when `format` is given the heuristic sniffing is bypassed -- callers that
9
+ // already know the format (e.g. impg emitting fasta-aln) can force it rather
10
+ // than rely on auto-detection, which is necessarily ambiguous between formats
11
+ // that share a leading '>' (fasta vs a3m)
12
+ export function parseMSA(text, currentAlignment = 0, format) {
13
+ if (format === 'stockholm') {
14
+ return new StockholmMSA(text, currentAlignment);
15
+ }
16
+ if (format === 'a3m') {
17
+ return new A3mMSA(text);
18
+ }
19
+ if (format === 'fasta') {
20
+ return new FastaMSA(text);
21
+ }
22
+ if (format === 'emf') {
23
+ return new EmfMSA(text);
24
+ }
25
+ if (format === 'clustal') {
26
+ return new ClustalMSA(text);
27
+ }
9
28
  if (stockholmSniff(text)) {
10
29
  return new StockholmMSA(text, currentAlignment);
11
30
  }
@@ -1 +1 @@
1
- {"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/msa/index.ts"],"names":[],"mappings":"AAAA,OAAO,MAAM,MAAM,aAAa,CAAA;AAChC,OAAO,UAAU,MAAM,iBAAiB,CAAA;AACxC,OAAO,MAAM,MAAM,aAAa,CAAA;AAChC,OAAO,QAAQ,MAAM,eAAe,CAAA;AACpC,OAAO,YAAY,EAAE,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAA;AAEhE,OAAO,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAA;AACrC,OAAO,EAAE,OAAO,IAAI,WAAW,EAAE,MAAM,kBAAkB,CAAA;AASzD,MAAM,UAAU,QAAQ,CAAC,IAAY,EAAE,gBAAgB,GAAG,CAAC;IACzD,IAAI,cAAc,CAAC,IAAI,CAAC,EAAE,CAAC;QACzB,OAAO,IAAI,YAAY,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAA;IACjD,CAAC;IACD,IAAI,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QACvB,OAAO,IAAI,MAAM,CAAC,IAAI,CAAC,CAAA;IACzB,CAAC;IACD,IAAI,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;QACzB,OAAO,IAAI,QAAQ,CAAC,IAAI,CAAC,CAAA;IAC3B,CAAC;IACD,IAAI,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3B,OAAO,IAAI,MAAM,CAAC,IAAI,CAAC,CAAA;IACzB,CAAC;IACD,OAAO,IAAI,UAAU,CAAC,IAAI,CAAC,CAAA;AAC7B,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,CAAC,UAAkB;IACpD,OAAO,UAAU,CAAC,UAAU,CAAC,OAAO,EAAE,EAAE,CAAC,CAAA;AAC3C,CAAC;AAED,OAAO,EAAE,OAAO,IAAI,MAAM,EAAE,MAAM,aAAa,CAAA;AAC/C,OAAO,EAAE,OAAO,IAAI,UAAU,EAAE,MAAM,iBAAiB,CAAA;AACvD,OAAO,EAAE,OAAO,IAAI,MAAM,EAAE,MAAM,aAAa,CAAA;AAC/C,OAAO,EAAE,OAAO,IAAI,QAAQ,EAAE,MAAM,eAAe,CAAA;AACnD,OAAO,EAAE,OAAO,IAAI,YAAY,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAA"}
1
+ {"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/msa/index.ts"],"names":[],"mappings":"AAAA,OAAO,MAAM,MAAM,aAAa,CAAA;AAChC,OAAO,UAAU,MAAM,iBAAiB,CAAA;AACxC,OAAO,MAAM,MAAM,aAAa,CAAA;AAChC,OAAO,QAAQ,MAAM,eAAe,CAAA;AACpC,OAAO,YAAY,EAAE,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAA;AAEhE,OAAO,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAA;AACrC,OAAO,EAAE,OAAO,IAAI,WAAW,EAAE,MAAM,kBAAkB,CAAA;AAWzD,4EAA4E;AAC5E,6EAA6E;AAC7E,8EAA8E;AAC9E,0CAA0C;AAC1C,MAAM,UAAU,QAAQ,CACtB,IAAY,EACZ,gBAAgB,GAAG,CAAC,EACpB,MAAkB;IAElB,IAAI,MAAM,KAAK,WAAW,EAAE,CAAC;QAC3B,OAAO,IAAI,YAAY,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAA;IACjD,CAAC;IACD,IAAI,MAAM,KAAK,KAAK,EAAE,CAAC;QACrB,OAAO,IAAI,MAAM,CAAC,IAAI,CAAC,CAAA;IACzB,CAAC;IACD,IAAI,MAAM,KAAK,OAAO,EAAE,CAAC;QACvB,OAAO,IAAI,QAAQ,CAAC,IAAI,CAAC,CAAA;IAC3B,CAAC;IACD,IAAI,MAAM,KAAK,KAAK,EAAE,CAAC;QACrB,OAAO,IAAI,MAAM,CAAC,IAAI,CAAC,CAAA;IACzB,CAAC;IACD,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QACzB,OAAO,IAAI,UAAU,CAAC,IAAI,CAAC,CAAA;IAC7B,CAAC;IACD,IAAI,cAAc,CAAC,IAAI,CAAC,EAAE,CAAC;QACzB,OAAO,IAAI,YAAY,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAA;IACjD,CAAC;IACD,IAAI,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QACvB,OAAO,IAAI,MAAM,CAAC,IAAI,CAAC,CAAA;IACzB,CAAC;IACD,IAAI,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;QACzB,OAAO,IAAI,QAAQ,CAAC,IAAI,CAAC,CAAA;IAC3B,CAAC;IACD,IAAI,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3B,OAAO,IAAI,MAAM,CAAC,IAAI,CAAC,CAAA;IACzB,CAAC;IACD,OAAO,IAAI,UAAU,CAAC,IAAI,CAAC,CAAA;AAC7B,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,CAAC,UAAkB;IACpD,OAAO,UAAU,CAAC,UAAU,CAAC,OAAO,EAAE,EAAE,CAAC,CAAA;AAC3C,CAAC;AAED,OAAO,EAAE,OAAO,IAAI,MAAM,EAAE,MAAM,aAAa,CAAA;AAC/C,OAAO,EAAE,OAAO,IAAI,UAAU,EAAE,MAAM,iBAAiB,CAAA;AACvD,OAAO,EAAE,OAAO,IAAI,MAAM,EAAE,MAAM,aAAa,CAAA;AAC/C,OAAO,EAAE,OAAO,IAAI,QAAQ,EAAE,MAAM,eAAe,CAAA;AACnD,OAAO,EAAE,OAAO,IAAI,YAAY,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAA"}
@@ -37,6 +37,22 @@ that doesn't match anything`;
37
37
  const msa = parseMSA(unknown);
38
38
  expect(msa).toBeInstanceOf(ClustalMSA);
39
39
  });
40
+ test('explicit format bypasses auto-detection', () => {
41
+ // a3m-looking input (lowercase inserts, differing row lengths) forced to
42
+ // FASTA, and FASTA-looking input forced to A3M
43
+ const a3mLooking = `>seq1
44
+ ACDEFghiKLMNPQ
45
+ >seq2
46
+ ACDEFKLMNPQ`;
47
+ expect(parseMSA(a3mLooking)).toBeInstanceOf(A3mMSA);
48
+ expect(parseMSA(a3mLooking, 0, 'fasta')).toBeInstanceOf(FastaMSA);
49
+ const fastaLooking = `>seq1
50
+ ACDEF
51
+ >seq2
52
+ GHIKL`;
53
+ expect(parseMSA(fastaLooking)).toBeInstanceOf(FastaMSA);
54
+ expect(parseMSA(fastaLooking, 0, 'a3m')).toBeInstanceOf(A3mMSA);
55
+ });
40
56
  });
41
57
  describe('FastaMSA colon-in-name lookup', () => {
42
58
  const fasta = `>EU105457.1|chr09:67680268..67675529_LTR/Copia
@@ -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;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,4BAA4B,EAAE,GAAG,EAAE;IAC1C,IAAI,CAAC,qDAAqD,EAAE,GAAG,EAAE;QAC/D,MAAM,GAAG,GAAG,IAAI,QAAQ,CAAC,sCAAsC,CAAC,CAAA;QAChE,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;QAChD,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAC1C,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"}
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;IAEF,IAAI,CAAC,yCAAyC,EAAE,GAAG,EAAE;QACnD,yEAAyE;QACzE,+CAA+C;QAC/C,MAAM,UAAU,GAAG;;;YAGX,CAAA;QACR,MAAM,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC,CAAC,cAAc,CAAC,MAAM,CAAC,CAAA;QACnD,MAAM,CAAC,QAAQ,CAAC,UAAU,EAAE,CAAC,EAAE,OAAO,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAA;QAEjE,MAAM,YAAY,GAAG;;;MAGnB,CAAA;QACF,MAAM,CAAC,QAAQ,CAAC,YAAY,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAA;QACvD,MAAM,CAAC,QAAQ,CAAC,YAAY,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,cAAc,CAAC,MAAM,CAAC,CAAA;IACjE,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,4BAA4B,EAAE,GAAG,EAAE;IAC1C,IAAI,CAAC,qDAAqD,EAAE,GAAG,EAAE;QAC/D,MAAM,GAAG,GAAG,IAAI,QAAQ,CAAC,sCAAsC,CAAC,CAAA;QAChE,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;QAChD,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAC1C,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/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "msa-parsers",
3
- "version": "5.3.0",
3
+ "version": "5.4.0",
4
4
  "license": "MIT",
5
5
  "type": "module",
6
6
  "repository": {
package/src/index.ts CHANGED
@@ -17,7 +17,7 @@ export {
17
17
  parseNewick,
18
18
  stockholmSniff,
19
19
  } from './msa/index.ts'
20
- export type { MSAParserType } from './msa/index.ts'
20
+ export type { MSAFormat, MSAParserType } from './msa/index.ts'
21
21
 
22
22
  // GFF parsing
23
23
  export {
package/src/msa/A3mMSA.ts CHANGED
@@ -87,6 +87,15 @@ export default class A3mMSA extends BaseMSA {
87
87
  return false
88
88
  }
89
89
 
90
+ // A true aligned FASTA (e.g. impg's fasta-aln output from soft-masked
91
+ // references) has all rows the same total length. A3M's defining feature is
92
+ // variable-length rows, since lowercase inserts live outside the fixed match
93
+ // grid. So equal-length rows are an alignment, not A3M, even when lowercase
94
+ // soft-masking is present.
95
+ if (new Set(seqs.map(s => s.length)).size === 1) {
96
+ return false
97
+ }
98
+
90
99
  let hasLowercase = false
91
100
  let firstMatchLen = -1
92
101
  let sameMatchLength = true
@@ -113,9 +113,19 @@ export default class StockholmMSA extends BaseMSA {
113
113
  id: 'secondaryStruct',
114
114
  name: 'Secondary-structure',
115
115
  data: this.secondaryStructureConsensus,
116
+ // WUSS notation pairs base-paired columns as open/close brackets of
117
+ // several nesting types (<>, (), [], {}); color all opens one way and
118
+ // closes the other so every helix (e.g. the tRNA acceptor stem, which
119
+ // is written with parens) is highlighted, not just the <> arms.
116
120
  customColorScheme: {
117
- '>': 'pink',
118
121
  '<': 'lightblue',
122
+ '(': 'lightblue',
123
+ '[': 'lightblue',
124
+ '{': 'lightblue',
125
+ '>': 'pink',
126
+ ')': 'pink',
127
+ ']': 'pink',
128
+ '}': 'pink',
119
129
  },
120
130
  },
121
131
  ]
@@ -0,0 +1,57 @@
1
+ import { describe, expect, test } from 'vitest'
2
+
3
+ import A3mMSA from './A3mMSA.ts'
4
+ import FastaMSA from './FastaMSA.ts'
5
+ import { parseMSA } from './index.ts'
6
+
7
+ // impg (https://github.com/pangenome/impg) pipes `impg query -o fasta-aln` into
8
+ // scripts/faln2html.py, which feeds the FASTA string to
9
+ // MSAModelF().create({ data: { msa } }). These tests pin the behavior that
10
+ // integration relies on.
11
+
12
+ describe('impg fasta-aln integration', () => {
13
+ // PanSN-style names + region coordinates with colons, dash gaps, uppercase
14
+ const block = `>HG002#1#chr1:1000-1040
15
+ ACGTACGT--ACGTNNNNACGTACGTACGTACGTACGT
16
+ >HG003#1#chr1:1000-1040
17
+ ACGTAC-TGGACGTNNNNACGTAC--ACGTACGTACGT
18
+ >CHM13#0#chr1:1000-1040
19
+ ACGTACGTGGACGTNNNNACGT----ACGTACGTACGT`
20
+
21
+ test('uppercase block is parsed as FASTA, not A3M/Clustal', () => {
22
+ const msa = parseMSA(block)
23
+ expect(msa).toBeInstanceOf(FastaMSA)
24
+ expect(msa.getNames()).toEqual([
25
+ 'HG002#1#chr1:1000-1040',
26
+ 'HG003#1#chr1:1000-1040',
27
+ 'CHM13#0#chr1:1000-1040',
28
+ ])
29
+ expect(msa.getWidth()).toBe(38)
30
+ })
31
+
32
+ test('rows are retrievable by their colon-containing region name', () => {
33
+ const msa = parseMSA(block)
34
+ expect(msa.getRow('HG002#1#chr1:1000-1040')).toBe(
35
+ 'ACGTACGT--ACGTNNNNACGTACGTACGTACGTACGT',
36
+ )
37
+ })
38
+
39
+ // If impg extracts from a soft-masked reference FASTA, aligned rows carry
40
+ // lowercase bases. A3M sniff runs before FASTA and triggers on
41
+ // hasLowercase && equal upper+dash match-length across rows -- which these
42
+ // rows satisfy by coincidence. The rows are all the same total length though,
43
+ // so they are a real alignment and must parse as FASTA, not A3M (A3M would
44
+ // strip the lowercase into separate insert columns and break alignment).
45
+ const softMasked = `>HG002#1#chr1:1000-1030
46
+ acgtACGTAC--GTACGTACGTACGTacgt
47
+ >HG003#1#chr1:1000-1030
48
+ ACGTacgtAC--GTACGTacgtACGTACGT`
49
+
50
+ test('soft-masked equal-length block parses as FASTA, not A3M', () => {
51
+ const msa = parseMSA(softMasked)
52
+ expect(msa).toBeInstanceOf(FastaMSA)
53
+ expect(msa).not.toBeInstanceOf(A3mMSA)
54
+ const rows = msa.getNames().map(n => msa.getRow(n))
55
+ expect(new Set(rows.map(r => r.length)).size).toBe(1)
56
+ })
57
+ })
@@ -46,6 +46,24 @@ that doesn't match anything`
46
46
 
47
47
  expect(msa).toBeInstanceOf(ClustalMSA)
48
48
  })
49
+
50
+ test('explicit format bypasses auto-detection', () => {
51
+ // a3m-looking input (lowercase inserts, differing row lengths) forced to
52
+ // FASTA, and FASTA-looking input forced to A3M
53
+ const a3mLooking = `>seq1
54
+ ACDEFghiKLMNPQ
55
+ >seq2
56
+ ACDEFKLMNPQ`
57
+ expect(parseMSA(a3mLooking)).toBeInstanceOf(A3mMSA)
58
+ expect(parseMSA(a3mLooking, 0, 'fasta')).toBeInstanceOf(FastaMSA)
59
+
60
+ const fastaLooking = `>seq1
61
+ ACDEF
62
+ >seq2
63
+ GHIKL`
64
+ expect(parseMSA(fastaLooking)).toBeInstanceOf(FastaMSA)
65
+ expect(parseMSA(fastaLooking, 0, 'a3m')).toBeInstanceOf(A3mMSA)
66
+ })
49
67
  })
50
68
 
51
69
  describe('FastaMSA colon-in-name lookup', () => {
package/src/msa/index.ts CHANGED
@@ -14,7 +14,32 @@ export type MSAParserType =
14
14
  | EmfMSA
15
15
  | ClustalMSA
16
16
 
17
- export function parseMSA(text: string, currentAlignment = 0): MSAParserType {
17
+ export type MSAFormat = 'stockholm' | 'a3m' | 'fasta' | 'emf' | 'clustal'
18
+
19
+ // when `format` is given the heuristic sniffing is bypassed -- callers that
20
+ // already know the format (e.g. impg emitting fasta-aln) can force it rather
21
+ // than rely on auto-detection, which is necessarily ambiguous between formats
22
+ // that share a leading '>' (fasta vs a3m)
23
+ export function parseMSA(
24
+ text: string,
25
+ currentAlignment = 0,
26
+ format?: MSAFormat,
27
+ ): MSAParserType {
28
+ if (format === 'stockholm') {
29
+ return new StockholmMSA(text, currentAlignment)
30
+ }
31
+ if (format === 'a3m') {
32
+ return new A3mMSA(text)
33
+ }
34
+ if (format === 'fasta') {
35
+ return new FastaMSA(text)
36
+ }
37
+ if (format === 'emf') {
38
+ return new EmfMSA(text)
39
+ }
40
+ if (format === 'clustal') {
41
+ return new ClustalMSA(text)
42
+ }
18
43
  if (stockholmSniff(text)) {
19
44
  return new StockholmMSA(text, currentAlignment)
20
45
  }