org.molgenis.data.annotation.core.filter

Class MultiAllelicResultFilter

java.lang.Object

org.molgenis.data.annotation.core.filter.MultiAllelicResultFilter

All Implemented Interfaces:

EntityProcessor, ResultFilter

public class MultiAllelicResultFilter
extends Object
implements ResultFilter

TODO: Support multi-allelic combination fields. These fields contain not only info for ref-alt pairs, but also values for all possible alt-alt pairs. For example, the "AC_Het" field in ExAC:

For 2 alt alleles, there are 3 values in the AC_Het field, the third being the count for T-A: 1 6536051 . C T,A 11129.51 PASS AC_Adj=5,3;AC_Het=5,3,0;AC_Hom=0,0; Alt-allele combinations in AC_Het field occur in the following order: C-T, C-A, T-A.

For 3 alt alleles, there are 6 values (3+2+1): 15 66641732 rs2063690 G C,A,T 35371281.87 PASS AC_Adj=13570,3,2;AC_Het=11380,1,2,2,0,0;AC_Hom=1094,0,0; Alt-allele combinations in AC_Het field occur in the following order: G-C, G-A, G-T, C-A, C-T, A-T.

For 4 alt alleles, there are 10 values (4+3+2+1): 21 45650009 rs3831401 T G,C,TG,A 8366813.26 PASS AC_Adj=2528,3415,1,0;AC_Het=934,1240,0,0,725,1,0,0,0,0;AC_Hom=434,725,0,0; Alt-allele combinations in AC_Het field occur in the following order: T-G, T-C, T-TG, T-A, G-C, G-TG, G-A, C-TG, C-A, TG-A.

TODO: Smart matching of non-obvious ref and alt alleles. Right now, we only 'string match' the exact values of ref and alt alleles. However, depending on which other genotypes you call for a certain genomic position, the same variant may be denoted in a different way. For example: "1 231094050 GA G" is the same variant as the GAA/GA in "1 231094050 GAA GA,G", but to allow notation of the the AA deletion (in GAA/G), the ref was written as GAA instead of GA.

This can be tricky. Consider this variant: 1 6529182 . TTCCTCC TTCC

And after some puzzling, you will find that it is seen in ExAC: 1 6529182 . TTCCTCCTCC TTCCTCC,TTCC,T,TTCCTCCTCCTCC,TTCCTCCTCCTCCTCC,TTCCTCCTCCTCCTCCTCCTCC

But here denoted as "TTCCTCCTCC/TTCCTCC". In both cases, a TCC was deleted, but in ExAC this variant is trailed with another TCC. Finding and parsing these variants to correctly match them against databases such as 1000 Genomes and ExAC would be very valuable.

Constructor Summary

Constructors

Constructor and Description
MultiAllelicResultFilter(List<Attribute> alleleSpecificAttributes, boolean mergeMultilineResourceResults, VcfAttributes vcfAttributes)
MultiAllelicResultFilter(List<Attribute> alleleSpecificAttributes, VcfAttributes vcfAttributes)

Method Summary

Modifier and Type	Method and Description
com.google.common.base.Optional<Entity>	filterResults(Iterable<Entity> resourceEntities, Entity sourceEntity, boolean updateMode)
Collection<Attribute>	getRequiredAttributes()
Iterable<Entity>	merge(Iterable<Entity> resourceEntities) Combine ALT information per reference allele (in VCF there is only 1 reference by letting ALT vary, but that might not always be the case)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

MultiAllelicResultFilter

public MultiAllelicResultFilter(List<Attribute> alleleSpecificAttributes,
                                boolean mergeMultilineResourceResults,
                                VcfAttributes vcfAttributes)

MultiAllelicResultFilter

public MultiAllelicResultFilter(List<Attribute> alleleSpecificAttributes,
                                VcfAttributes vcfAttributes)

Method Detail

getRequiredAttributes

public Collection<Attribute> getRequiredAttributes()

Specified by:

getRequiredAttributes in interface EntityProcessor

Returns:

Collection of atomic Attribute that must be present in the annotated Entity.

filterResults

public com.google.common.base.Optional<Entity> filterResults(Iterable<Entity> resourceEntities,
                                                             Entity sourceEntity,
                                                             boolean updateMode)

Specified by:

filterResults in interface ResultFilter

merge

public Iterable<Entity> merge(Iterable<Entity> resourceEntities)

Combine ALT information per reference allele (in VCF there is only 1 reference by letting ALT vary, but that might not always be the case)

So we want to support this hypothetical example: 3 300 G A 0.2|23.1 3 300 G T -2.4|0.123 3 300 G X -0.002|2.3 3 300 G C 0.5|14.5 3 300 GC A 0.2|23.1 3 300 GC T -2.4|0.123 3 300 C GX -0.002|2.3 3 300 C GC 0.5|14.5

So we want to support this hypothetical example: 3 300 G A 0.2|23.1 3 300 G T -2.4|0.123 3 300 G X -0.002|2.3 3 300 G C 0.5|14.5 3 300 GC A 0.2|23.1 3 300 GC T -2.4|0.123 3 300 C GX -0.002|2.3 3 300 C GC 0.5|14.5

and it should become:

3 300 G A,T,X,C 0.2|23.1,-2.4|0.123,-0.002|2.3,0.5|14.5 3 300 GC A,T 0.2|23.1,-2.4|0.123 3 300 C GX,GC -0.002|2.3,0.5|14.5

3 300 G A,T,X,C 0.2|23.1,-2.4|0.123,-0.002|2.3,0.5|14.5 3 300 GC A,T 0.2|23.1,-2.4|0.123 3 300 C GX,GC -0.002|2.3,0.5|14.5

so that the multi-allelic filter can then find back the appropriate values as if it were a multi-allelic VCF line