@jbrowse/cigar-utils 5.0.0-beta.11 → 5.0.0-beta.14

package/README.md

@@ -77,7 +77,7 @@ segments — without paying to split and rejoin the tag around the filter.
 
 ```js
 // type signature
-(entries: string[], id: string, strand: number | undefined, readName: string | undefined, normalize?: boolean | undefined) => {…}[]
+(entries: string[], id: string, strand: number | undefined, readName: string | undefined, normalize?: boolean | undefined) => { refName: string; start: number; end: number; seqLength: number; clipLengthAtStartOfRead: number; CIGAR: string; strand: 1 | -1; uniqueId: string; mate: { start: number; end: number; refName: string | undefined; }; }[]
 ```
 
 [Source code](https://github.com/GMOD/jbrowse-components/blob/main/packages/cigar-utils/src/mismatchParser.ts)

package/esm/cigarCategoryBits.d.ts

@@ -0,0 +1,19 @@
+/** A CIGAR carries a reference skip, the `N` a spliced aligner writes across an intron. */
+export declare const CIGAR_HAS_SKIP = 2;
+/** A CIGAR carries a deletion of {@link SV_INDEL_MIN_BP} or more. */
+export declare const CIGAR_HAS_SV_DELETION = 4;
+/** A CIGAR carries an insertion of {@link SV_INDEL_MIN_BP} or more. */
+export declare const CIGAR_HAS_SV_INSERTION = 8;
+/** The indel length from which callers file an indel as a structural variant. */
+export declare const SV_INDEL_MIN_BP = 50;
+/**
+ * {@link CIGAR_HAS_SKIP}, {@link CIGAR_HAS_SV_DELETION} and
+ * {@link CIGAR_HAS_SV_INSERTION} over a packed CIGAR (`len << 4 | op`).
+ */
+export declare function numericCigarCategoryBits(cigar: ArrayLike<number>): number;
+/**
+ * {@link numericCigarCategoryBits} over a CIGAR string, walked without
+ * parsing it into an array, since a synteny block's CIGAR can run to millions
+ * of operations.
+ */
+export declare function cigarStringCategoryBits(cigar: string): number;

package/esm/cigarCategoryBits.js

@@ -0,0 +1,87 @@
+import { CIGAR_D, CIGAR_I, CIGAR_N } from './cigarConstants.js';
+/** A CIGAR carries a reference skip, the `N` a spliced aligner writes across an intron. */
+export const CIGAR_HAS_SKIP = 2;
+/** A CIGAR carries a deletion of {@link SV_INDEL_MIN_BP} or more. */
+export const CIGAR_HAS_SV_DELETION = 4;
+/** A CIGAR carries an insertion of {@link SV_INDEL_MIN_BP} or more. */
+export const CIGAR_HAS_SV_INSERTION = 8;
+/** The indel length from which callers file an indel as a structural variant. */
+export const SV_INDEL_MIN_BP = 50;
+const SV_INDEL_MIN_PACKED = SV_INDEL_MIN_BP << 4;
+// The op count from which the branch-free scan wins. Interleaved in one
+// process (`plugins/alignments/benches/categoryBits.bench.ts`), the branching
+// form took 11.6 ms over 153,677 short reads against 13.5 branch-free, and
+// 15.4 ms over 335 long reads of 7,300 ops against 9.7: a long read's ops mix
+// in an order no predictor learns, and a short read has too few to repay the
+// arithmetic.
+const BRANCH_FREE_MIN_OPS = 32;
+/**
+ * {@link CIGAR_HAS_SKIP}, {@link CIGAR_HAS_SV_DELETION} and
+ * {@link CIGAR_HAS_SV_INSERTION} over a packed CIGAR (`len << 4 | op`).
+ */
+export function numericCigarCategoryBits(cigar) {
+    if (cigar.length < BRANCH_FREE_MIN_OPS) {
+        let bits = 0;
+        for (let i = 0; i < cigar.length; i++) {
+            const packed = cigar[i];
+            const op = packed & 0xf;
+            if (op === CIGAR_N) {
+                bits |= CIGAR_HAS_SKIP;
+            }
+            else if (packed >= SV_INDEL_MIN_PACKED) {
+                if (op === CIGAR_D) {
+                    bits |= CIGAR_HAS_SV_DELETION;
+                }
+                else if (op === CIGAR_I) {
+                    bits |= CIGAR_HAS_SV_INSERTION;
+                }
+            }
+        }
+        return bits;
+    }
+    // One bit per op kind seen, and per op kind seen at SV length
+    let seen = 0;
+    let svLength = 0;
+    for (let i = 0; i < cigar.length; i++) {
+        const packed = cigar[i];
+        const op = packed & 0xf;
+        seen |= 1 << op;
+        svLength |= +(packed >= SV_INDEL_MIN_PACKED) << op;
+    }
+    return ((seen & (1 << CIGAR_N) ? CIGAR_HAS_SKIP : 0) |
+        (svLength & (1 << CIGAR_D) ? CIGAR_HAS_SV_DELETION : 0) |
+        (svLength & (1 << CIGAR_I) ? CIGAR_HAS_SV_INSERTION : 0));
+}
+const CHAR_D = 68;
+const CHAR_I = 73;
+const CHAR_N = 78;
+/**
+ * {@link numericCigarCategoryBits} over a CIGAR string, walked without
+ * parsing it into an array, since a synteny block's CIGAR can run to millions
+ * of operations.
+ */
+export function cigarStringCategoryBits(cigar) {
+    let bits = 0;
+    let len = 0;
+    for (let i = 0; i < cigar.length; i++) {
+        const code = cigar.charCodeAt(i);
+        if (code >= 48 && code <= 57) {
+            len = len * 10 + code - 48;
+        }
+        else {
+            if (code === CHAR_N) {
+                bits |= CIGAR_HAS_SKIP;
+            }
+            else if (len >= SV_INDEL_MIN_BP) {
+                if (code === CHAR_D) {
+                    bits |= CIGAR_HAS_SV_DELETION;
+                }
+                else if (code === CHAR_I) {
+                    bits |= CIGAR_HAS_SV_INSERTION;
+                }
+            }
+            len = 0;
+        }
+    }
+    return bits;
+}

package/esm/index.d.ts

@@ -18,7 +18,7 @@ export { mdToMismatches2 } from './mdToMismatches2.ts';
 export { getNextRefPos } from './getNextRefPos.ts';
 export { refWindowToRead } from './refWindowToRead.ts';
 export type { CigarCursor, ReadWindow } from './refWindowToRead.ts';
-export { numericCigarHasSkip } from './numericCigarHasSkip.ts';
+export { CIGAR_HAS_SKIP, CIGAR_HAS_SV_DELETION, CIGAR_HAS_SV_INSERTION, SV_INDEL_MIN_BP, cigarStringCategoryBits, numericCigarCategoryBits, } from './cigarCategoryBits.ts';
 export { numericCigarToString } from './numericCigarToString.ts';
 export { connectionEndpointBps, readLeadingBodyDir, readLeadingBp, readTrailingBodyDir, readTrailingBp, } from './readEndpoints.ts';
 export { clipLengthAtStartOfReadNumeric, featurizeSA, featurizeSAEntries, getClip, getLength, getLengthOnRef, getLengthSansClipping, getMismatches, parseCigar2, parseCigar2Typed, parseCigar, splitSA, } from './mismatchParser.ts';

package/esm/index.js

@@ -18,7 +18,7 @@ export { cigarToMismatches2 } from './cigarToMismatches2.js';
 export { mdToMismatches2 } from './mdToMismatches2.js';
 export { getNextRefPos } from './getNextRefPos.js';
 export { refWindowToRead } from './refWindowToRead.js';
-export { numericCigarHasSkip } from './numericCigarHasSkip.js';
+export { CIGAR_HAS_SKIP, CIGAR_HAS_SV_DELETION, CIGAR_HAS_SV_INSERTION, SV_INDEL_MIN_BP, cigarStringCategoryBits, numericCigarCategoryBits, } from './cigarCategoryBits.js';
 export { numericCigarToString } from './numericCigarToString.js';
 export { connectionEndpointBps, readLeadingBodyDir, readLeadingBp, readTrailingBodyDir, readTrailingBp, } from './readEndpoints.js';
 export { clipLengthAtStartOfReadNumeric, featurizeSA, featurizeSAEntries, getClip, getLength, getLengthOnRef, getLengthSansClipping, getMismatches, parseCigar2, parseCigar2Typed, parseCigar, splitSA, } from './mismatchParser.js';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jbrowse/cigar-utils",
-  "version": "5.0.0-beta.11",
+  "version": "5.0.0-beta.14",
   "type": "module",
   "description": "Pure CIGAR / MD / mismatch parsers and types — no rendering or framework deps",
   "keywords": [

package/esm/numericCigarHasSkip.d.ts

@@ -1 +0,0 @@
-export declare function numericCigarHasSkip(cigar: ArrayLike<number>): boolean;

package/esm/numericCigarHasSkip.js

@@ -1,11 +0,0 @@
-import { CIGAR_N } from './cigarConstants.js';
-// Whether a packed CIGAR (`len << 4 | op`) carries a reference skip — the `N`
-// a spliced aligner writes across an intron.
-export function numericCigarHasSkip(cigar) {
-    for (let i = 0; i < cigar.length; i++) {
-        if ((cigar[i] & 0xf) === CIGAR_N) {
-            return true;
-        }
-    }
-    return false;
-}