molstar
Version:
A comprehensive macromolecular library.
117 lines (116 loc) • 4.7 kB
JavaScript
;
/**
* Copyright (c) 2020-2022 mol* contributors, licensed under MIT, See LICENSE file for more info.
*
* @author David Sehnal <david.sehnal@gmail.com>
*/
Object.defineProperty(exports, "__esModule", { value: true });
exports.computeDensityIsocontourValues = exports.createSphericalCollocationDensityGrid = void 0;
var tslib_1 = require("tslib");
var util_1 = require("../../mol-data/util");
var grid3d_1 = require("../../mol-gl/compute/grid3d");
var mol_task_1 = require("../../mol-task");
var debug_1 = require("../../mol-util/debug");
var data_model_1 = require("./data-model");
var compute_1 = require("./gpu/compute");
function createSphericalCollocationDensityGrid(params, orbitals, webgl) {
var _this = this;
return mol_task_1.Task.create('Spherical Collocation Grid', function (ctx) { return tslib_1.__awaiter(_this, void 0, void 0, function () {
var cubeGrid, matrix, grid, isovalues;
return tslib_1.__generator(this, function (_a) {
switch (_a.label) {
case 0:
cubeGrid = (0, data_model_1.initCubeGrid)(params);
if (!(0, grid3d_1.canComputeGrid3dOnGPU)(webgl)) return [3 /*break*/, 2];
if (debug_1.isTimingMode)
webgl.timer.mark('createSphericalCollocationDensityGrid');
return [4 /*yield*/, (0, compute_1.gpuComputeAlphaOrbitalsDensityGridValues)(ctx, webgl, cubeGrid, orbitals)];
case 1:
matrix = _a.sent();
if (debug_1.isTimingMode)
webgl.timer.markEnd('createSphericalCollocationDensityGrid');
return [3 /*break*/, 3];
case 2: throw new Error('Missing OES_texture_float WebGL extension.');
case 3:
grid = (0, data_model_1.createGrid)(cubeGrid, matrix, [0, 1, 2]);
if (!params.doNotComputeIsovalues) {
isovalues = computeDensityIsocontourValues(matrix, 0.85);
}
return [2 /*return*/, { grid: grid, isovalues: isovalues }];
}
});
}); });
}
exports.createSphericalCollocationDensityGrid = createSphericalCollocationDensityGrid;
function computeDensityIsocontourValues(input, cumulativeThreshold) {
var weightSum = 0;
for (var i = 0, _i = input.length; i < _i; i++) {
var v = input[i];
var w = Math.abs(v);
weightSum += w;
}
var avgWeight = weightSum / input.length;
var minWeight = 3 * avgWeight;
// do not try to identify isovalues for degenerate data
// e.g. all values are almost same
if (Math.abs(avgWeight - input[0] * input[0]) < 1e-5) {
return { negative: void 0, positive: void 0 };
}
var size = 0;
while (true) {
var csum = 0;
size = 0;
for (var i = 0, _i = input.length; i < _i; i++) {
var v = input[i];
var w = Math.abs(v);
if (w >= minWeight) {
csum += w;
size++;
}
}
if (csum / weightSum > cumulativeThreshold) {
break;
}
minWeight -= avgWeight;
}
var values = new Float32Array(size);
var weights = new Float32Array(size);
var indices = new Int32Array(size);
var o = 0;
for (var i = 0, _i = input.length; i < _i; i++) {
var v = input[i];
var w = Math.abs(v);
if (w >= minWeight) {
values[o] = v;
weights[o] = w;
indices[o] = o;
o++;
}
}
(0, util_1.sortArray)(indices, function (indices, i, j) { return weights[indices[j]] - weights[indices[i]]; });
var cweight = 0, cutoffIndex = 0;
for (var i = 0; i < size; i++) {
cweight += weights[indices[i]];
if (cweight / weightSum >= cumulativeThreshold) {
cutoffIndex = i;
break;
}
}
var positive = Number.POSITIVE_INFINITY, negative = Number.NEGATIVE_INFINITY;
for (var i = 0; i < cutoffIndex; i++) {
var v = values[indices[i]];
if (v > 0) {
if (v < positive)
positive = v;
}
else if (v < 0) {
if (v > negative)
negative = v;
}
}
return {
negative: negative !== Number.NEGATIVE_INFINITY ? negative : void 0,
positive: positive !== Number.POSITIVE_INFINITY ? positive : void 0,
};
}
exports.computeDensityIsocontourValues = computeDensityIsocontourValues;