parser.js

import yaml from "js-yaml";
import fs from "fs";
import * as utils from "./utils.js";
import { crossReferencesDict } from "./identifiers.js";

const getInfoFromYaml = (yamlFile) => {
  // extract information from the YAML file
  const [metadata, metabolites, reactions, genes, compartments] = yaml.load(
    fs.readFileSync(yamlFile, "utf8")
  );
  const metadataSection = metadata.metaData || metadata.metadata;
  const model = utils.toLabelCase(metadataSection.short_name);
  const version = `V${metadataSection.version.replace(/\./g, "_")}`;
  const isHuman = metadataSection.short_name === "Human-GEM";
  return [
    metadata,
    metabolites,
    reactions,
    genes,
    compartments,
    metadataSection,
    model,
    version,
    isHuman,
  ];
};

const getComponentSvgRel = (
  content,
  component,
  svgNodes,
  modelDir,
  metadataSection
) => {
  // get SVG files for compartments and subsystems
  const filename = `${component}SVG.tsv`;
  const mappingFile = utils.getFile(modelDir, filename);
  const isCustom = component === "custom";

  let svgRels = [];
  if (mappingFile) {
    const lines = fs
      .readFileSync(mappingFile, { encoding: "utf8", flag: "r" })
      .split("\n")
      .filter(Boolean);
    const filenameSet = new Set(); // check uniqness of values in the file
    for (let i = 0; i < lines.length; i++) {
      if (lines[i][0] == "#" || lines[i][0] == "@") {
        continue;
      }

      let componentName, mapName, mapFilename;

      const columns = lines[i].split("\t").map((e) => e.trim());
      if (isCustom) {
        [mapName, mapFilename] = columns;
      } else {
        [componentName, mapName, mapFilename] = columns;
      }

      if (
        componentName &&
        !content[component].map((e) => e.name).includes(componentName)
      ) {
        throw new Error(
          `${component} "${componentName}" does not exist in the model "${metadataSection.short_name}"`
        );
      }

      if (filenameSet.has(mapFilename)) {
        throw new Error(
          `map ${mapFilename} can only be linked to one ${component} in the model "${metadataSection.short_name}"`
        );
      }
      filenameSet.add(mapFilename);

      if (!/^[a-z0-9_]+[.]svg$/.test(mapFilename)) {
        throw new Error(
          `map "${mapFilename}" referenced by ${metadataSection.short_name}/${filename} is invalid`
        );
      }
      svgNodes.push({
        id: mapFilename.split(".")[0],
        filename: mapFilename,
        customName: mapName,
      });

      if (componentName) {
        svgRels.push({
          [`${component}Id`]: utils.idfyString(componentName),
          svgMapId: mapFilename.split(".")[0],
        });
      }
    }
  } else {
    console.log(
      `Warning: cannot find mappingfile ${filename} in path`,
      modelDir
    );
  }

  return svgRels;
};

const getPMIDs = (PMIDSset, componentIdDict) => {
  // get pubmed IDs from the componentIdDict, which is read from the YAML file
  const reactionPMID = [];
  const PMIDs = [];
  for (const reactionId in componentIdDict.reaction) {
    const ECList = componentIdDict.reaction[reactionId].ec;
    let PMIDList = componentIdDict.reaction[reactionId].references;
    if (PMIDList) {
      PMIDList.split(";").forEach((pubmedReferenceId) => {
        pubmedReferenceId = pubmedReferenceId.trim();
        if (pubmedReferenceId.match("^PMID")) {
          pubmedReferenceId = pubmedReferenceId.replace(/PMID:*/g, "").trim();
          const isnum = /^\d+$/.test(pubmedReferenceId);
          if (isnum) {
            reactionPMID.push({ reactionId, pubmedReferenceId });
            if (!PMIDSset.has(pubmedReferenceId)) {
              PMIDs.push(pubmedReferenceId);
              PMIDSset.add(pubmedReferenceId);
            }
          }
        }
      });
    }
  }
  return [PMIDs, reactionPMID];
};

const getGeneAnnotation = (componentIdDict, modelDir) => {
  // get annotaitons for genes from the genes tsv file
  const geneAnnoFile = utils.getFile(modelDir, /genes[.]tsv$/);
  if (!geneAnnoFile) {
    console.log(
      "Warning: cannot find gene annotation file genes.tsv in path",
      modelDir
    );
  } else {
    // TODO use one of the csv parsing lib (sync)
    const lines = fs
      .readFileSync(geneAnnoFile, { encoding: "utf8", flag: "r" })
      .split("\n")
      .filter(Boolean);
    for (let i = 0; i < lines.length; i++) {
      if (lines[i][0] == "#" || lines[i][0] == "@") {
        continue;
      }
      // thefunction, ec and catalytic_activity are not defined in the new TSV
      // format and thus set the default value as empty
      const thefunction = "";
      const ec = "";
      const catalytic_activity = "";
      const [
        geneId,
        geneENSTID,
        geneENSPID,
        geneUniProtID,
        name,
        geneEntrezID,
        alternateName,
        synonyms,
      ] = lines[i].split("\t").map((e) => utils.trim(e, '"'));
      if (geneId in componentIdDict.gene) {
        //only keep the ones in the model
        const gene = componentIdDict.gene[geneId];
        Object.assign(gene, {
          name,
          alternateName,
          synonyms,
          function: thefunction,
        }); // other props are not in the db design, TODO remove them?
      }
    }
  }
};

const getComponentExternalDb = (
  externalIdNodes,
  externalIdDBMap,
  extNodeIdTracker,
  component,
  componentIdDict,
  modelDir
) => {
  // get externalId from the components tsv files
  const externalIdDBComponentRel = [];
  const filename = `${component}s.tsv`;
  const extIDFile = utils.getFile(modelDir, filename);
  const fcomponent =
    component === "metabolite" ? "compartmentalizedMetabolite" : component;

  if (extIDFile) {
    // TODO use one of the csv parsing lib (sync)
    const lines = fs
      .readFileSync(extIDFile, { encoding: "utf8", flag: "r" })
      .split("\n")
      .filter(Boolean);

    var headerArr = [];
    var contentArr = [];
    for (let i = 0; i < lines.length; i++) {
      if (lines[i][0] == "#") {
        continue;
      } else if (i == 0) {
        /*read the header line*/
        headerArr = lines[i].split("\t").map((e) => e.trim());
        continue;
      } else {
        contentArr = lines[i].split("\t").map((e) => utils.trim(e, '"'));
      }

      const id = contentArr[0];
      if (!(id in componentIdDict[fcomponent])) {
        //only keep the ones in the model
        console.log(
          "Warning: id " +
            id +
            " not in " +
            " componentIdDict[" +
            fcomponent +
            "]"
        );
        continue;
      }

      if (fcomponent == "gene") {
        /*add two more items Ensembl and Protein Atlas which is not included in the new format*/
        headerArr.push("geneEnsemblID");
        headerArr.push("geneProteinAtlasID");
        contentArr.push(id); /*For Ensembl, externalId is equal to id*/
        contentArr.push(id); /*For Protein Atlas, externalId is equal to id*/
      }
      const numItem = contentArr.length;

      for (let j = 1; j < numItem; j++) {
        const header = headerArr[j];
        const regexGene = "gene.*ID$";
        const regexRxn = "rxn.*ID$";
        const regexMet = "met.*ID$";
        if (
          (fcomponent == "gene" && header.match(regexGene) == null) ||
          (fcomponent == "reaction" && header.match(regexRxn) == null) ||
          (fcomponent == "compartmentalizedMetabolite" &&
            header.match(regexMet) == null)
        ) {
          continue;
        }
        const crossReferencesArray = Object.values(crossReferencesDict);
        const referenceData = crossReferencesArray.find((item) =>
          item.headers.includes(header)
        );
        const dbName = referenceData ? referenceData.db : "";
        const dbPrefix = referenceData ? referenceData.dbPrefix : "";
        const suffix =
          fcomponent == "gene"
            ? referenceData.geneSuffix
            : fcomponent == "reaction"
            ? referenceData.reactionSuffix
            : referenceData.compoundSuffix;
        const rawExternalId = utils.cleanExternalId(contentArr[j], dbName);
        if (rawExternalId == "") {
          //ignore the record whithout any valid externalId
          continue;
        }
        // There might be multiple ids in one externalId item
        const externalIdArr = rawExternalId.split(";").map((e) => e.trim());

        for (var externalId of externalIdArr) {
          var url = "";
          if (dbPrefix != "" && externalId != "") {
            const urlMap = `https://identifiers.org/${dbPrefix}${suffix}`;
            url = `${urlMap}:${externalId}`;
          }

          const externalDbEntryKey = `${dbName}${externalId}${url}`; // diff url leads to new nodes!
          externalId = dbName === "ChEBI" ? "CHEBI:" + externalId : externalId;

          let node = null;
          if (externalDbEntryKey in externalIdDBMap) {
            node = externalIdDBMap[externalDbEntryKey]; // reuse the node and id
          } else {
            node = { id: extNodeIdTracker, dbName, externalId, url };
            externalIdDBMap[externalDbEntryKey] = node;
            extNodeIdTracker += 1;

            // save the node for externalDBs.csv
            externalIdNodes.push(node);
          }

          // save the relationships between the node and the current component ID (reaction, gene, etc)
          externalIdDBComponentRel.push({ id, externalDbId: node.id }); // e.g. geneId, externalDbId
        }
      }
    }
  } else {
    console.log(
      `Warning: cannot find external ID file ${filename} in path`,
      modelDir
    );
  }
  return [extNodeIdTracker, fcomponent, externalIdDBComponentRel];
};

export {
  getInfoFromYaml,
  getComponentSvgRel,
  getPMIDs,
  getGeneAnnotation,
  getComponentExternalDb,
};