@jbrowse/core/util/cluster.js

JBrowse 2 core libraries used by plugins

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.euclideanDistance = euclideanDistance;
exports.averageDistance = averageDistance;
exports.clusterData = clusterData;
const stopToken_1 = require("@jbrowse/core/util/stopToken");
function toP(n) {
    return Number.parseFloat((n * 100).toFixed(1));
}
function euclideanDistance(a, b) {
    const size = Math.min(a.length, b.length);
    let sum = 0;
    for (let index = 0; index < size; index++) {
        sum += (a[index] - b[index]) * (a[index] - b[index]);
    }
    return Math.sqrt(sum);
}
function averageDistance(setA, setB, distances) {
    let distance = 0;
    const lenA = setA.length;
    const lenB = setB.length;
    for (let i = 0; i < lenA; i++) {
        for (let j = 0; j < lenB; j++) {
            distance += distances[setA[i]][setB[j]];
        }
    }
    return distance / setA.length / setB.length;
}
function clusterData({ data, distance = euclideanDistance, linkage = averageDistance, onProgress, stopToken, }) {
    let start = performance.now();
    const distances = [];
    for (let i = 0; i < data.length; i++) {
        if (performance.now() - start > 400) {
            (0, stopToken_1.checkStopToken)(stopToken);
            start = performance.now();
        }
        if (onProgress) {
            onProgress(`Making distance matrix: ${toP(i / (data.length - 1))}%`);
        }
        const row = [];
        for (let j = 0; j < data.length; j++) {
            row.push(distance(data[i], data[j]));
        }
        distances.push(row);
    }
    const clusters = data.map((_datum, index) => ({
        height: 0,
        indexes: [Number(index)],
    }));
    let clustersGivenK = [];
    start = performance.now();
    for (let iteration = 0; iteration < data.length; iteration++) {
        if (performance.now() - start > 400) {
            (0, stopToken_1.checkStopToken)(stopToken);
            start = performance.now();
        }
        if (onProgress) {
            onProgress(`Clustering: ${toP((iteration + 1) / data.length)}%`);
        }
        clustersGivenK.push(clusters.map(cluster => cluster.indexes));
        if (iteration >= data.length - 1) {
            break;
        }
        let nearestDistance = Infinity;
        let nearestRow = 0;
        let nearestCol = 0;
        for (let row = 0; row < clusters.length; row++) {
            for (let col = row + 1; col < clusters.length; col++) {
                const distance = linkage(clusters[row].indexes, clusters[col].indexes, distances);
                if (distance < nearestDistance) {
                    nearestDistance = distance;
                    nearestRow = row;
                    nearestCol = col;
                }
            }
        }
        const newCluster = {
            indexes: [
                ...clusters[nearestRow].indexes,
                ...clusters[nearestCol].indexes,
            ],
            height: nearestDistance,
            children: [clusters[nearestRow], clusters[nearestCol]],
        };
        clusters.splice(Math.max(nearestRow, nearestCol), 1);
        clusters.splice(Math.min(nearestRow, nearestCol), 1);
        clusters.push(newCluster);
    }
    clustersGivenK = [[], ...clustersGivenK.reverse()];
    return {
        clusters: clusters[0],
        distances: distances,
        order: clusters[0].indexes,
        clustersGivenK: clustersGivenK,
    };
}