The more consistent data set is now available ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/20101123/interim_phase1_release/
The genotypes here are complete and phased for all 1094 individuals
1000 Genomes: Phased Or Not?
Q1. Running
zcat ALL.chrX.BI_Beagle.20100804.genotypes.vcf.gz | grep -v ^## | cut -f 345 | cut -d ':' -f 1 | grep -v '\./\.' | grep -v '|' | head
yields
NA18981
0/0
0/0
1/1
0/0
0/0
0/0
0/0
0/0
0/0
while running
zcat ALL.chrX.BI_Beagle.20100804.genotypes.vcf.gz | grep -v ^## | cut -f 345 | cut -d ':' -f 1 | grep -v '\./\.' | head
yields
NA18981
0|0
0|0
0|0
0|0
0|0
0|0
0|0
1|0
0|0
Is NA18981 phased, or is it not? Is it partially phased? If yes - then what rule/convention explains this partiality? (I know that microsatellite calls are unphased in phased genomes, but I believe I haven't seen any in this file.)
Q2. For somatic chromosomes (I've only checked this on chrs 1 and 2, but I assume this pattern is characteristic for all autosomal chromosomes) all 629 samples appear to be phased - that is, their genotypes at all positions are either unknown ./. or phased (e.g. 0|1). So are all 629 samples really phased on all somatic chromosomes?
Somewhat related: http://biostar.stackexchange.com/questions/5315/phased-and-unphased-genotypes-in-vcf-files-does-the-order-of-alleles-matter
1 answer
You should not trust the X chromosome genotype calls. They were not made in the proper way. On autosomes, the majority of SNPs are arbitrarily phased, but some are unphased due to call set merging. The upcoming release will be much cleaner.
As to phasing itself, you can always arbitrarily phase heterozygotes. The question is how many switch errors we make. The answer to this question is largely unknown on a data set like 1000g.
Log in to answer this question.
More posts like this
-
Bwa Run With Fastq File With Different Read Length
written by ttom
-
Haplotype Frequency Calculator For Vcf Files
written by diviya.smith •
-
1000G Query / Using Tabix With A Proxy
written by secretjess
-
Command To Change Fasta File Format
written by biolab
-
Discrepancy Between Ncbi / Vcf And Ncbi / Html
written by Pierre Lindenbaum
-
Formatting Beadstudio Final Report Into Plink
written by romsen •
-
Samtools To Analyze Bwa Output
written by Dawnoflife •
-
Why Total Exon Length Of 2/3Rds Of Ucsc Knowngenes Is Non-Divisible By 3?
written by Chronos
-
How To Decompress 1000Genomes Bgzip-Compressed Files Using Java
written by Pierre Lindenbaum
-
Phased And Unphased Genotypes In Vcf Files: Does The Order Of Alleles Matter?
written by Chronos
You should check the documentation of the programs used to phase the data. They certainly contain a section about phasing chr X and haploids. Chr X is hemizygous in certain positions. Phasing can be problematic.
They used Beagle (judging from the chrX filename), and I'll have to read its manual sooner or later. Jarretinha, did you mean to say that 1000 genomes project considers pseudo-autosomal segments of Y as diploid segments on the corresponding chrX coordinates? That would make sense (as there is no Y chromosome anywhere in the data), but then... Why there are no non-diploid genotypes on chrX? (They should have no slash/pipe in them, just a number or a dot.)
Please be more descriptive!
+1 I actually think the question is very clear.
ok, the question is very clear now after the edit. I've removed the -1.