Changes

From Genome Analysis Wiki
Jump to navigationJump to search

Variant Normalization (view source)

Revision as of 09:55, 1 August 2016

631 bytes added ,  09:55, 1 August 2016
→‎Other versions
Line 104: Line 104:     
Thus A and B have to be at the same position and have the same length and variant normalization is unique.
 
Thus A and B have to be at the same position and have the same length and variant normalization is unique.
−
  −
= I can find an example where the normalization algorithm fails =
  −
  −
Hi Terry,
  −
  −
Thanks for the report. This is an interesting example.
  −
  −
But before I begin, I would like to distinguish the difference between normalization and decomposition of variants (as we defined it)
  −
  −
Decomposition of variants involves the breaking down of a variant record into multiple records. It may be done vertically - as in multiallelics becoming biallelics or it can be done horizontally - a cluster of indels and SNPs represented as a complex variant being splitted up into several records. Horizontal decompositions in general do not have a unique solution.
  −
  −
Normalization involves ensuring the representation of a variant record is left aligned and parsimonious and does not increase or decrease the number of records representing that variant. Normalization can be applied to biallelic variants or multiallelic variants. The problem of normalization is solvable and there exists a unique representation that is left aligned and parsimonious. Mathematical proof is published. [http://bioinformatics.oxfordjournals.org/content/early/2015/03/22/bioinformatics.btv112]
  −
  −
Supporting haplotype reconstruction is actually not the goal of vt's normalization, it is meant for allowing one to compare the alleles of variant call sets from different variant callers applied possibly to multiple samples.
  −
  −
The notion of normalization that you described involves reconstruction of haplotypes, and you are right to say that there should be some inbuilt collision detection mechanism. It was a really nice example and in the context of a single sample and the assumption that the alternate alleles must occur on the same haplotype, it is correct.
  −
      
= Implementation =
 
= Implementation =
Line 234: Line 217:  
*GATK v3.1-1-g07a4bf8
 
*GATK v3.1-1-g07a4bf8
 
*vt normalize v0.5
 
*vt normalize v0.5
 +
 +
= Here is an example where this normalization algorithm fails =
 +
 +
We distinguish the concepts of normalization and decomposition/reconstruction of variants as follows:
 +
 +
 +
Normalization involves reducing representations of a variant to a canonical representation. Normalization can be applied to biallelic variants or multiallelic variants. The problem of normalization is solvable and there exists a unique representation that is left aligned and parsimonious. Mathematical proof is published. [http://bioinformatics.oxfordjournals.org/content/suppl/2015/02/19/btv112.DC1/VtNormApplicationNote_supp_20141113_1346.pdf]
 +
 +
 +
Decomposition of variants involves the breaking down of a variant record into multiple records. It may be done vertically - as in multiallelics becoming biallelics or it can be done horizontally - a cluster of indels and SNPs represented as a complex variant being splitted up into several records. Horizontal decompositions in general do not have a unique solution.  Similarly, reconstruction combines several variant records into a single record and can be done vertically and horizontally too. Vertical decomposition of a multiallelic variant to a set of biallelic records is a many to one function.  Construction of a set of biallelic variants into a multiallelic record is not unique as you need to considered all possible permutations of the haplotypes containing your alleles. 
 +
 +
 +
If your example contains the decomposition or reconstruction of variants, then it is probable that you can find inconsistencies. 
 +
 +
 +
It is important to distinguish the difference between normalization and decomposition/reconstruction.  The notion of normalization implies that a variant can be reduced to a standardized form.  If you were to include decomposition and reconstruction in your notion of normalization, you are  bound to find inconsistencies simply due to the inherent issues of identifiability. 
 +
 +
 +
When performing decomposition and construction, I think the following factors should be considered:
 +
 +
* Are your variants describing just a single individual or a population?
 +
* Are the genotypes (if any) in your individual(s) phased?
 +
 +
Depending on the context, you will obtain different answers.
 +
 +
[https://github.com/atks/vt/issues/16 An example of inconsistent variant representation due to using vt normalize]
    
= Citation =
 
= Citation =
   −
[http://bioinformatics.oxfordjournals.org/content/early/2015/02/19/bioinformatics.btv112.abstract?keytype=ref&ijkey=2kB1TkBGzkoP1gd Adrian Tan, Gonçalo R. Abecasis and Hyun Min Kang. (2015)  Unified Representation of Genetic Variants. Bioinformatics. doi: 10.1093/bioinformatics/btv112 ]
+
[http://bioinformatics.oxfordjournals.org/content/31/13/2202 Adrian Tan, Gonçalo R. Abecasis and Hyun Min Kang. (2015)  Unified Representation of Genetic Variants. Bioinformatics.]
 +
 
 +
= Translations =
 +
 
 +
A mandarin translation can be found [http://www.lyon0804.com/fan-yi-variant-normalization.html here]
    
= Maintained by =
 
= Maintained by =
    
This page is maintained by  [mailto:atks@umich.edu Adrian].
 
This page is maintained by  [mailto:atks@umich.edu Adrian].
Atks
1,102

edits

Retrieved from "http://genome.sph.umich.edu/wiki/Special:MobileDiff/13074...14371"

Navigation menu