uniprotParser/index.ts

// https://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/taxonomic_divisions/uniprot_sprot_human.xml.gz

import fs from 'fs'
import readline from 'readline'
import { XMLParser } from 'fast-xml-parser'
import jsonata from 'jsonata'

const line$ = (path: string) => readline.createInterface({
    input: fs.createReadStream(path),
    crlfDelay: Infinity
})

const makeIndex = async (filePath: string, indexPath?: string) => {
    indexPath = indexPath || filePath + '.jsi'

    let byteAcc = 0
    const fromSel = new RegExp("^<entry")
    const toSel = new RegExp("^</entry>")
    const valSel = new RegExp('<accession>')
    let tmp =  {values:[]} as any
    for await (const line of line$(filePath)) {
        if(fromSel.test(line)) tmp['from'] = byteAcc

        byteAcc += (line.length + 1)

        if (valSel.test(line)) tmp['values'].push(line.match("<accession>(.*?)</accession>")![1]) // 'uck 

        if(toSel.test(line)) {
            await fs.promises.appendFile(indexPath, tmp.values.join(';') + '\t' + tmp.from + '\t' + byteAcc + '\n')
            tmp = {values:[]}
        }
    }
}

const readOffset = (path: string, from:number, to:number) => {
    return new Promise<string>(async (resolve, reject) => {
        const size = to - from
        const buffer = Buffer.alloc(size);
        let filehandle
        try {
            filehandle = await fs.promises.open(path, 'r+');
            await filehandle.read(buffer, 0, buffer.length, from);
        } finally {
            if (filehandle) {
                await filehandle.close()
                resolve(buffer.toString())
            }
        }
    })
}

const getEntryOffset = async (dbPath:string, accession:string): Promise<number[]> => {
    const indexPath = dbPath + '.jsi'
    if (!fs.existsSync(indexPath)) await makeIndex(dbPath)
    const lineSel = new RegExp(accession)
    for await (const line of line$(indexPath)) {
        if (lineSel.test(line)) return [Number(line.split('\t')[1]),Number(line.split('\t')[2])]
    }
    return [0, 0]
}

const getEnrty = async (dbPath:string, accession:string) => {
    const parser = new XMLParser({
        ignoreAttributes: false, 
        alwaysCreateTextNode: false, 
        attributeNamePrefix: "",
        textNodeName: "value",
        allowBooleanAttributes: true,
    })

    const offsets = await getEntryOffset(dbPath, accession)
    return parser.parse(await readOffset(dbPath, offsets[0], offsets[1]))
}

const getEntryFromGeneName = async (idmappingPath: string, dbPath:string, geneName:string) => {
    const sel = new RegExp('Gene_Name\t' + geneName)
    let accessions = []
    for await (const line of line$(idmappingPath)) {
        if(sel.test(line)) accessions.push(line.split('\t')[0])
    }
    return await getEnrty(dbPath, accessions[0])   
}

const getInteractionsFromEntry = async (json:any) => {
    const uniprotIDs = Array.isArray(json.entry.accession) ? json.entry.accession : [json.entry.accession]
    
    // Comment interactant
    const res_inter = jsonata(`entry.comment[type="interaction"].interactant`).evaluate(json)
    let genes_interactant:any[] = []
    if (res_inter) {
        genes_interactant = [...new Set(
            res_inter
            .filter((e:any) => !uniprotIDs.includes(e.label))
            .map((e:any) => e.label)
            .filter((e:any) => e)
        )]
    }
    
    // Reference scope = INTERACTION WITH
    const scope_inter = jsonata(`entry.reference[scope ~> /INTERACTION WITH/i ]`).evaluate(json)
    let genes_scope_inter = []
    if (scope_inter) {
        const comment_scope_inters = Array.isArray(scope_inter) ? scope_inter : [scope_inter]
        const comment_scope_inters_genes = comment_scope_inters.map((e:any) => ({
            txt: Array.isArray(e.scope) ? e.scope.filter((ee:any) => ee.match(/INTERACTION\ WITH/)).join() : e.scope,
            ...e
        })).map((e:any) => ({
            interaction: e.txt.substring(e.txt.indexOf("INTERACTION WITH ") + "INTERACTION WITH ".length, e.txt.length),
            ...e
        }))
        genes_scope_inter = [...new Set(comment_scope_inters_genes.map((e:any) => e.interaction))].flatMap((e:any) => e.split(/; | AND /))
    }
    
    // Comment subunit
    const comment_subunit = jsonata(`entry.comment[type="subunit"].text.value`).evaluate(json)
    let comment_subunits_genes = []
    if (comment_subunit) {
        const comment_subunits = Array.isArray(comment_subunit) ? comment_subunit : [comment_subunit]
        comment_subunits_genes = comment_subunits
        .flatMap((e:any) => e.replace(/ *\([^)]*\) */g, '').split(/\n/))
        .filter((e:any) => /Interacts/.test(e))
        .flatMap((e:any) => e.match(/[A-Z][A-Z0-9]{2,}/g))
    }

    let res = [...new Set([...genes_scope_inter, ...genes_interactant, ...comment_subunits_genes])].sort().filter((e:any) => typeof e === 'string').filter(_=>_)
    
    if(res.length > 0) res = res.flatMap((e:any) => e.match(/[A-Z]{2,}[A-Z0-9]{1,}/g)).filter(_=>_)
    
    return res
}

// const findDistance = async (idmappingPath: string, dbPath:string, geneNameA:string, geneNameB:string, maxDistance = 6) => {
//     let rounds = [[geneNameA]]

//     let tree = {[geneNameA]: {}} as {[key:string]:any}
//     let run = true

//     let a = tree
//     Object.keys(a).map((gene) => )

//     let nIter = 0
//     while(nIter <= maxDistance && run) {
//         for (const gA of rounds[nIter]) {
//             console.log(nIter,gA);
            
//             const tmp = await getInteractionsFromEntry(await getEntryFromGeneName(idmappingPath, dbPath, gA))
//             if (tmp.includes(geneNameB)) { run = false; break }
//             rounds.push(tmp)
//         }
//         nIter++
//     }

//     //console.log(rounds);
    
    
//     return nIter
// }

export { makeIndex, readOffset, getEnrty, getEntryFromGeneName, getInteractionsFromEntry }