Skip to content

Annotation hell

May 12, 2015

I’ve been trying to sort out strand alignment for our data, which is typed on the Illumina Omni 2.5M platform(s).  We have two sets of data – some is typed on the ‘octo’ chip (HumanOmni2.5-8v1_A manifest) while one cohort is typed on the ‘quad’ chip (HumanOmni2.5-4v1_D manifest).  To avoid downstream issues with strand alignment, we’ve taken to aligning and flipping data to match the forward strand of the reference sequence up-front.  To do this we’ve used a consensus between the Illumina manifest strand and the strand files provided by Will Rayner, which are based on blatting the probes to the reference sequence.

Unfortunately stuff goes wrong:


This plot shows allele frequencies on the two chips at well-typed SNPs – and looks bonkers.  Turns out that a subset of SNPs in HumanOmni2.5-4v1_D have the alleles swapped.  You can detect these by comparing the SNP column of the manifest (which lists the alleles) with the last base in the allele probe sequences.  For 18763 variants, the bases don’t match – these are coloured red above.  (And it’s not just me that thinks so: these same 18763 variants have got the probe sequences listed the other way round in the HumanOmni2.5-4v1_H manifest.)

A search for the same problem in other manifests for this chip – I looked at HumanOmni2.5-8v1_A, HumanOmni2.5-4v1_B (which is mapped to build 36) and HumanOmni2.5-4v1_H – turns up no issues of the same kind*, i.e. where both probe sequences are listed in the manifest file, the SNP alleles always match the last base of the probe sequences in those manifests.  Plotting allele frequency in the aligned data for that chip against frequency in similar 1000 Genomes populations confirms things look ok.

*Curiously, in the Human1M-Duov3_C manifest, which is for a different chip entirely, a single variant has a mismatch – rs4865393 is listed as A/G but has probes for A/T.  dbSNP lists this SNP as triallelic.

No comments yet

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: