import { AlphaFoldPayload } from '@nightingale-elements/nightingale-structure'; // from color scale B:0,H:0.1132,V:0.2264,L:0.3395,A:0.4527,l:0.5895,h:0.7264,p:0.8632,P:1 const certainlyBenign = 0; const benign = 0.1132; const veryLikelyBenign = 0.2264; const likelyBenign = 0.3395; const ambiguous = 0.4527; const likelyAmbiguous = 0.5895; const likelyPathogenic = 0.7264; const pathogenic = 0.8632; const certainlyPathogenic = 1; export const rowSplitter = /\s*\n\s*/; export const cellSplitter = /^(.)(\d+)(.),(.+),(\w+)$/; const pathogenicityCategories = [ { min: certainlyBenign, max: benign, code: 'H' }, { min: benign, max: veryLikelyBenign, code: 'V' }, { min: veryLikelyBenign, max: likelyBenign, code: 'L' }, { min: likelyBenign, max: ambiguous, code: 'A' }, { min: ambiguous, max: likelyAmbiguous, code: 'l' }, { min: likelyAmbiguous, max: likelyPathogenic, code: 'h' }, { min: likelyPathogenic, max: pathogenic, code: 'p' }, { min: pathogenic, max: certainlyPathogenic, code: 'P' }, ]; const getPathogenicityCode = (score: number): string | undefined => { for (const { min, max, code } of pathogenicityCategories) { if (score >= min && score < max) { return code; } } }; type Row = { wildType: string; position: number; mutated: string; pathogenicityScore: number; pathogenicityLabel: string; }; const parseCSV = (rawText: string): string => { const positions: Array> = []; for (const [i, row] of rawText.split(rowSplitter).entries()) { if (i === 0 || !row) { continue; } const match = row.match(cellSplitter); if (!match) { continue; } const [ , wildType, positionString, mutated, pathogenicityScore, pathogenicityLabel, ] = match; const position = +positionString; const index = position - 1; if (!positions[index]) { positions[index] = []; } positions[index].push({ wildType, position, mutated, pathogenicityScore: +pathogenicityScore, pathogenicityLabel, }); } const out = []; for (const position of positions) { // maximum // const value = Math.max( // ...position.map((variation) => variation.pathogenicityScore) // ); // average const value = position.reduce( (acc, variation) => acc + +variation.pathogenicityScore, 0 ) / position.length; const letter = getPathogenicityCode(value); out.push(letter); } return out.join(''); }; // Load and parse const loadAndParseAnnotations = async ( url: string ): Promise => { try { const payload = await fetch(url); const rawCSV = await payload.text(); return parseCSV(rawCSV); } catch (e) { console.error('Could not load AlphaMissense pathogenicity score', e); return undefined; } }; type PartialProtein = { sequence: { sequence: string; }; }; const transformData = async ( data: AlphaFoldPayload, protein: PartialProtein ) => { const alphaFoldSequenceMatch = data?.filter( ({ sequence, amAnnotationsUrl }) => protein.sequence.sequence === sequence && amAnnotationsUrl ); if (alphaFoldSequenceMatch.length === 1) { const url = alphaFoldSequenceMatch[0].amAnnotationsUrl; if (!url) return undefined; const heatmapData = await loadAndParseAnnotations(url); return heatmapData; } else if (alphaFoldSequenceMatch.length > 1) { console.warn( `Found more than one matches (${alphaFoldSequenceMatch.length}) for AlphaMissense pathogenicity score against protein sequence: ${protein.sequence}` ); } }; export default transformData;