Version: Next

Id Resolvers

The ID resolver uses the files in the specified directory to create a large map. The key for the map is the unique identifier (the MOD ID, for example the MGI:, RGD, FBgn, ZFIN: identifiers). The values in the map are all the symbols, old identifiers, dbxrefs (e.g. Ensembl).

unique gene identifier	symbol, name, ensembl ID …
MGI:97490	pax6, paired box gene 6 …

The ID resolver then uses this map to replace old or non-unique identifiers with the unique identifier. This allows genes to be merged correctly into the database, and lets each mine be interoperable with other friendly mines.

The ID resolver is used in several data sources, Homologene for example.

If you look at the Homologene data, you'll see they don't use the MGI identifier. See:


1212	10090	18508	Pax6	7305369 NP_038655.1
1212	10116	25509	Pax6	6981334 NP_037133.1

When parsing the Homologene data file, the ID resolver replaces the symbol "Pax6" with the MGI identifier. The parser sets MGI:97490 to be the primary identifier then stores the gene to the database. Similarly, it replaces Pax6 with "RGD:3258" for the rat gene. And so on.

ID resolvers available in InterMine#


EntrezGeneIdResolverFactory	NCBI gene info for a collection of organisms	ftp://ftp.ncbi.nih.gov/gene/DATA/gene_info.gz
FlyBaseIdResolverFactory	flybase chado db, for ‘’D.melanogaster’’ only	ftp://ftp.flybase.net/releases/current/psql flybase chado
WormBaseChadoIdResolverFactory	wormbase chado db, for ‘’C.elegans’’ only	modENCODE specific
ZfinIdentifiersResolverFactory	zebrafish ids	http://zfin.org/downloads/identifiersForIntermine.txt
MgiIdentifiersResolverFactory	mouse ids	ftp://ftp.informatics.jax.org/pub/reports/MRK_List2.rpt
RgdIdentifiersResolverFactory	rat ids	ftp://rgd.mcw.edu/pub/data_release/GENES_RAT.txt
HgncIdResolverFactory	HGNC human gene ids	Uses the biomart service at http://www.genenames.org
EnsemblIdResolverFactory	Ensembl id	customised
HumanIdResolverFactory	human ids	customised

Using ID Resolvers in InterMine data converters#

Many data converters use the Entrez (NCBI) Gene ID resolver:

Download the identifier file -
ftp://ftp.ncbi.nih.gov/gene/DATA/gene_info.gz
Unzip the file to /DATA_DIR/ncbi/gene_info
Create a sub directory /DATA_DIR/idresolver/ as file root path and a symbolic link entrez to the file
$ cd /DATA_DIR/idresolver/
$ ln -s /DATA_DIR/ncbi/gene_info entrez
Add the root path to the file in ~/.intermine/MINE.properties
resolver.file.rootpath=/DATA_DIR/idresolver/

Id resolvers and corresponding symbolic to data file:

Resolver	Symbolic link
EntrezGeneIdResolverFactory	entrez
WormBaseChadoIdResolverFactory	wormid
ZfinIdentifiersResolverFactory	zfin
MgiIdentifiersResolverFactory	mgi
RgdIdentifiersResolverFactory	rgd
HgncIdResolverFactory	hgnc
EnsemblIdResolverFactory	ensembl
HumanIdResolverFactory	humangene

In the data converter, the ID resolver is given an identifier. The resolver then looks in the map for the identifier.

number of matches	returns
0	NULL
1	new identifier
>1	NULL

Using ID Resolvers in your data converters#

A factory will find data root path from ~/.intermine/MINE_NAME.properties, path needs to be absolute.

resolver.file.rootpath=/DATA_DIR/idresolver/

the key and the symbolic link of the data file need to be hard-coded in factory class, e.g. in EntrezGeneIdResolverFactory

private final String propKey = "resolver.file.rootpath";
private final String resolverFileSymbo = "entrez";

As for database case, e.g. flybase chado

# chado DB for flybase data

db.flybase.datasource.class=org.postgresql.jdbc3.Jdbc3PoolingDataSource
db.flybase.datasource.dataSourceName=db.flybase
db.flybase.datasource.serverName=NAME
db.flybase.datasource.databaseName=DBNAME
db.flybase.datasource.user=USER
db.flybase.datasource.password=PWD
db.flybase.datasource.maxConnections=10
db.flybase.driver=org.postgresql.Driver
db.flybase.platform=PostgreSQL

the key also needs to be hard-coded in factory class, e.g. in FlyBaseIdResolverFactory

private final String propName = "db.flybase";

Configuration#

The Entrez gene identifier source has a configuration file, entrezIdResolver_config.properties. You shouldn't have to edit this file.

This config will parse fruit fly identifiers, e.g. FLYBASE:FBgn0088803

7227.primaryIdentifier.xref=FLYBASE

If you don't want to strip the prefix from the identifier, use this config:

10116.primaryIdentifier.prefix=RGD:
10090.primaryIdentifier.prefix=MGI:

Warning The EBI changed how they format their data. If you have a recent data file, you do NOT want the above configuration for MGI.

To replace a taxonomy identifier with a strain, use the following:

4932.strains=559292

To ignore certain organisms, do this:

taxon.ignored = 7165,6239

IdResolverService#

IdResolverService is a java class providing static methods to get id resolver directly. It's also the most straight forward way to create an id resolver. For example, to create a fish id resolver by taxon id in a converter:

IdResolver rslvr = IdResolverService.getIdResolverByOrganism("7955");

You can use the IdResolverService to create resolver by taxon id, a list of taxon ids, or by organism, e.g.

IdResolver flyRslvr = IdResolverService.getFlyIdResolver();

Resolve an Id#

As the resolver maintains java maps of one or more organisms' identifiers, you must explicitly tell it which organism you want it to resolve for, e.g.

String pid = flyRslvr.resolveId(taxonId, identifier).iterator().next();

It is also possible there are two or more matching primary identifiers for a particular identifier, in this case, discard this identifier, e.g.

int resCount = flyRslvr.countResolutions(taxonId, identifier);
if (resCount  = 1) {
  LOG.info("RESOLVER: failed to resolve fly gene to one identifier, ignoring gene: "
          + identifier + " count: " + resCount + " FBgn: "
          + flyRslvr.resolveId(taxonId, identifier));
  return null;
}

Writing a New ID resolver#

An IdResolver factory will create an IdResolver which will read and parse data from a file or database containing identifier information, to save them to a Java map which will be written to a cached file.

The new factory class needs to inherit super class IdResolverFactory:

public class HumanIdResolverFactory extends IdResolverFactory

createIdResolver method:

// 1. check if the resolver which has the taxon and class has already been created
resolver.hasTaxonAndClassName(taxonId, this.clsCol.iterator().next())

// 2. Restore cached data from file. New data will be append to the cached file.
boolean isCachedIdResolverRestored = restoreFromFile(); 

// 3. To implement reading and parsing data from a customized file/db, see createFromFile method and createFromDb method.

createFromFile method:

// Ref HumanIdResolverFactory.java
// Parse a tab delimited file. Add to resolver.
String symbol = line[0];

resolver.addMainIds(taxonId, symbol, Collections.singleton(symbol));

createFromDb method:

// Ref FlyBaseIdResolverFactory.java
// 1. Set db connection parameters in MINE.properties, scroll up to see flybase chado setting.
// 2. Connect to the database and query the data.
// 3. Parse ResultSet, addIdsFromResultSet method

Multiple taxon ids:

// Ref EntrezGeneIdResolverFactory.java
public IdResolver getIdResolver(Set<String> taxonIds) {
      if (taxonIds == null || taxonIds.isEmpty()) {
          return null;
      }
      return getIdResolver(taxonIds, true);
}

Multiple classes:

// Ref FlyBaseIdResolverFactory.java
public FlyBaseIdResolverFactory(Set<String> clsCol) {
    // clsCol is set in parent class IdResolverFactory.java  
    this.clsCol = clsCol;
}

Multiple files or mixture of file and db:

// We don't have an example to handle muliple files, but one can always add them and parse them one by one.
// We have an example of handling db and file together, ref WormBaseIdResolverFactory.java

Add resolver factory to IdResolverService:

// Ref IdResolverService.java
public static IdResolver getHumanIdResolver() {
    return new HumanIdResolverFactory().getIdResolver(false);
}

public static IdResolver getHumanIdResolver(boolean failOnError) {
    return new HumanIdResolverFactory().getIdResolver(failOnError);
}

Future Plans#

generalized resolver factory which will read a configuration file to be aware of identifier information by column. e.g. type=tab, column.0=mainId, etc.