PDA

View Full Version : HLA types ?



thetick
02-15-2016, 07:33 PM
Has anyone found a list or preferably a tool to HLA determine types using 23andme data?
For example the MHC class I, II and III define at https://en.wikipedia.org/wiki/Human_leukocyte_antigen ?

Also has anyone compiled research on these types to determine likelihood of ancestry or medical predispositions?

I have used getmatch and R analysis in the region (location 29677984 and ends at location 33485635) but I wanted to know if more details currently exist?

warwick
02-15-2016, 08:23 PM
Has anyone found a list or preferably a tool to HLA determine types using 23andme data?
For example the MHC class I, II and III define at https://en.wikipedia.org/wiki/Human_leukocyte_antigen ?

Also has anyone compiled research on these types to determine likelihood of ancestry or medical predispositions?

I have used getmatch and R analysis in the region (location 29677984 and ends at location 33485635) but I wanted to know if more details currently exist?

My understanding is that it is challenging to get complete HLA data from the 23andme SNP chip results. You could probably get it either from a specialized panel or via WGS or exome data.

One data source for HLA freqs:
http://allelefrequencies.net

Example of one genotyping method for HLA:
http://www.illumina.com/products/trusight-hla-sequencing-panel.html

Disclosure of conflict of interest:
I am a FGC team member (CEO).

warwick
02-15-2016, 08:35 PM
My understanding is that it is challenging to get complete HLA data from the 23andme SNP chip results. You could probably get it either from a specialized panel or via WGS or exome data.

One data source for HLA freqs:
http://allelefrequencies.net



Example of one genotyping method for HLA:
http://www.illumina.com/products/trusight-hla-sequencing-panel.html

Disclosure of conflict of interest:
I am a FGC team member (CEO).

Vadim Verenich also discusses these issues here as well:
http://magnusducatus.blogspot.com/2011/12/experimental-test-inferring-hla.html


Disclosure of conflict of interest:
I am a FGC team member (CEO).

C J Wyatt III
02-15-2016, 08:39 PM
I am a FGC team member (CEO).

What is FGC?

Thanks,

Jack

warwick
02-15-2016, 08:49 PM
What is FGC?

Thanks,

Jack

See:
www.fullgenomes.com

thetick
02-15-2016, 09:19 PM
Disclosure of conflict of interest:
I am a FGC team member (CEO).

Thanks for reply as the MDLP link pretty much answers my question. Also thank you for your efforts with FGC. I'm not a current customer but plan to be one in the near future as the 30x FG price comes down below a grand.

Ann Turner
02-16-2016, 03:23 PM
It's difficult even with WGS / exome data. This article describes a method for achieving higher accuracy. But also notice how long it took to run the algorithm (11 hours for one sample and 17 hours for another sample).

http://biorxiv.org/content/early/2015/12/24/035253

AJL
02-16-2016, 03:47 PM
Has anyone found a list or preferably a tool to HLA determine types using 23andme data?

This was something I was quite interested in back in the DNA-Forums days.

As someone on the Canadian donor registry, I was able to supply Vadim with my HLA serotypes (obtained via an access to information request), as well as my 23andme and Family Finder genotypes.

The imputation tool (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3065693/) he ran matched the top-level groups in my case, even giving more precise subgroups than those obtained by serotyping. But as Ann mentioned, I believe it required heavy computer use to run.

I then ran the haplotypes through allelefrequencies. Interestingly one of my haplotypes -- my maternal, obviously -- occurs uncommonly in northern England but has a more Arctic distribution, occurring in Saami but peaking among Aleuts. In the end my overall histocompatibility, taking into account both sides, seems to be highest with Finns.

Since many Europeans score ~1% Arctic component, I guess I just happen to have inherited Arctic-style HLAs.

thetick
02-16-2016, 08:36 PM
It's difficult even with WGS / exome data. This article describes a method for achieving higher accuracy. But also notice how long it took to run the algorithm (11 hours for one sample and 17 hours for another sample).

http://biorxiv.org/content/early/2015/12/24/035253

Thanks for the reference, Ann. I'm not too concerned about the compute time as new software can almost always be improved drastically. Also by the time I have FG 30X state of the art computers will be 10X faster anyway. In a funny somewhat related note, at my current job as an application development , support and test role just last week I made a very simple configuration change during some standard software quality testing and found one of the processes that was taking 8 hours can now run in under 4 minutes. Functional but unoptimized software can have huge performance bottlenecks.

Now I'm not saying that will happen for HLA analysis, but there is always a chance a new or optimized algorithm will be an order of magnitude faster.

thetick
02-16-2016, 08:50 PM
This was something I was quite interested in back in the DNA-Forums days.

As someone on the Canadian donor registry, I was able to supply Vadim with my HLA serotypes (obtained via an access to information request), as well as my 23andme and Family Finder genotypes.

The imputation tool (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3065693/) he ran matched the top-level groups in my case, even giving more precise subgroups than those obtained by serotyping. But as Ann mentioned, I believe it required heavy computer use to run.

I then ran the haplotypes through allelefrequencies. Interestingly one of my haplotypes -- my maternal, obviously -- occurs uncommonly in northern England but has a more Arctic distribution, occurring in Saami but peaking among Aleuts. In the end my overall histocompatibility, taking into account both sides, seems to be highest with Finns.

Since many Europeans score ~1% Arctic component, I guess I just happen to have inherited Arctic-style HLAs.

https://oxfordhla.well.ox.ac.uk/hla -- Affymetrix has exclusively licensed the HLA*IMP algorithm from Peptide Groove and provides HLA*IMP-based HLA type imputation for Affymetrix arrays and other platforms.

Argh.. anyway thanks for all the details you provided. I will not have any time soon to review all this, but at a later point I may. And well when I have 30X results lying around I'll have incentive. Glad to see you found some very interesting details on your HLA. I think my HLA will be interesting too as the admixture tools all point to a Jewish or Middle Eastern affinity.