Converts Second Genome data files to a uBiome taxonomy file.

AV15-xxxx.otuTable.tsv

convert.js

"use strict";

var csv = require("ya-csv");

loadTaxa("stooldb_taxonomy_v1.0.tsv")
	.then(taxa => processSamples(process.openStdin(), taxa))
	.then(toUBiome)
	.then(printJson);

function loadTaxa(path) {
	return new Promise((resolve, reject) => {
		let taxa = {};
		let reader = csv.createCsvFileReader(path, { columnsFromHeader : true, "separator" : "\t" });
		reader.on("data", (row) => {
			let id = row[""];
			let taxon = {};
			for (let field in row) {
				if (field) {
					let value = row[field].replace(/^.__/, "");
					if (value !== "unclassified") {
						taxon[field.toLowerCase()] = value;
					}
				}
			}
			taxa[id] = taxon;
		});
		reader.on("end", () => {
			resolve(taxa);
		});
	});
}

function processSamples(stream, taxa) {
	return new Promise((resolve, reject) => {
		let samples = new Map();
		let reader = csv.createCsvStreamReader(process.openStdin(), { "separator" : "\t" });
		reader.on("data", (row) => {
			let id = row[0];
			let count = Number(row[2]);
			let taxon = taxa[id];
			for (let rank in taxon) {
				let key = rank + ":" + taxon[rank];
				samples.set(key, (samples.get(key) || 0) + count);
			}
		});
		reader.on("end", () => {
			resolve(samples);
		});
	});
}

function toUBiome(samples) {
	let counts = []
	let total = Math.max.apply(null, Array.from(samples.values()));
	for (let entry of samples) {
		let key = entry[0].split(":");
		let value = entry[1];
		counts.push({
			"count" : value,
			"count_norm" : Math.round(1000000 * value / total),
			"tax_rank" : key[0],
			"tax_name" : key[1]
		});
	}
	return Promise.resolve({ "ubiome_bacteriacounts" : counts });
}

function printJson(value) {
	console.log(JSON.stringify(value, null, 4));
}

stooldb_taxonomy_v1.0.tsv