@gmod/vcf/esm/Variant.js

High performance streaming Variant Call Format (VCF) parser in pure JavaScript

import { parseGenotypesOnly } from "./parseGenotypesOnly.js";
import { parseInfo } from "./parseInfo.js";
import { processGenotypes } from "./processGenotypes.js";
export class Variant {
    CHROM;
    POS;
    ID;
    REF;
    ALT;
    QUAL;
    FILTER;
    INFO;
    FORMAT;
    formatMeta;
    rest;
    sampleNames;
    constructor(line, infoMeta, formatMeta, sampleNames, strict) {
        const lineLen = line.length;
        let currChar = 0;
        let tabCount = 0;
        while (currChar < lineLen && tabCount < 9) {
            if (line.charCodeAt(currChar) === 9) {
                tabCount += 1;
            }
            currChar += 1;
        }
        const splitPos = tabCount === 9 ? currChar - 1 : currChar;
        const fields = line.slice(0, splitPos).split('\t');
        const rest = line.slice(splitPos + 1);
        const [CHROM, POS, ID, REF, ALT, QUAL, FILTER] = fields;
        const filter = FILTER === '.' ? undefined : FILTER?.split(';');
        if (strict && !fields[7]) {
            throw new Error("no INFO field specified, must contain at least a '.' (turn off strict mode to allow)");
        }
        this.CHROM = CHROM;
        this.POS = POS !== undefined ? +POS : 0;
        this.ID = ID === '.' ? undefined : ID?.split(';');
        this.REF = REF;
        this.ALT = ALT === '.' ? undefined : ALT?.split(',');
        this.QUAL = QUAL === undefined || QUAL === '.' ? undefined : +QUAL;
        this.FILTER = filter?.length === 1 && filter[0] === 'PASS' ? 'PASS' : filter;
        this.INFO =
            fields[7] === undefined || fields[7] === '.'
                ? {}
                : parseInfo(fields[7], infoMeta);
        this.FORMAT = fields[8];
        this.formatMeta = formatMeta;
        this.rest = rest;
        this.sampleNames = sampleNames;
    }
    SAMPLES() {
        const genotypes = {};
        const format = this.FORMAT;
        if (format) {
            const rest = this.rest.split('\t');
            const formatKeys = format.split(':');
            const isNumberType = formatKeys.map(k => {
                const t = this.formatMeta[k]?.Type;
                return t === 'Integer' || t === 'Float';
            });
            const numKeys = formatKeys.length;
            const samplesLen = this.sampleNames.length;
            for (let i = 0; i < samplesLen; i++) {
                const sample = this.sampleNames[i] ?? '';
                const sampleData = {};
                const sampleStr = rest[i] ?? '';
                const sampleStrLen = sampleStr.length;
                let colStart = 0;
                let colIdx = 0;
                for (let j = 0; j <= sampleStrLen; j++) {
                    if (j === sampleStrLen || sampleStr[j] === ':') {
                        const key = formatKeys[colIdx] ?? '';
                        const val = sampleStr.slice(colStart, j);
                        if (val === '' || val === '.') {
                            sampleData[key] = undefined;
                        }
                        else {
                            const items = val.split(',');
                            const itemsLen = items.length;
                            const isNum = isNumberType[colIdx];
                            const result = [];
                            for (let k = 0; k < itemsLen; k++) {
                                const ent = items[k] ?? '';
                                result.push(ent === '.' ? undefined : isNum ? +ent : ent);
                            }
                            sampleData[key] = result;
                        }
                        colStart = j + 1;
                        colIdx += 1;
                        if (colIdx >= numKeys) {
                            break;
                        }
                    }
                }
                genotypes[sample] = sampleData;
            }
        }
        return genotypes;
    }
    GENOTYPES() {
        return parseGenotypesOnly(this.FORMAT ?? '', this.rest, this.sampleNames);
    }
    processGenotypes(callback) {
        processGenotypes(this.FORMAT ?? '', this.rest, this.sampleNames.length, callback);
    }
    toJSON() {
        return {
            CHROM: this.CHROM,
            POS: this.POS,
            ID: this.ID,
            REF: this.REF,
            ALT: this.ALT,
            QUAL: this.QUAL,
            FILTER: this.FILTER,
            INFO: this.INFO,
            FORMAT: this.FORMAT,
        };
    }
}
//# sourceMappingURL=Variant.js.map