What Would You Do With Your Personal Genome Data?

Question

[Mary]

I was wrestling with this question myself some time ago, but I was just spurred to re-visit it by David Ewing Duncan, Experimental Man. He's just received his data from Complete Genomics, and now he asks for suggestions on what to do:

This is an appeal: Send me you ideas for how best to interpret my newly sequenced complete genome!

I think this group is better positioned than most to have this discussion. Some time ago I answered this for myself, and I'll summarize that here and add a couple of things. But I'm also wondering what you guys would do with your data. And I'll point David to this thread in the comments at his blog.

1) Assessment and QC: checking the files, formats, etc, look at sequences compared to reference genome and well-known genes, and to other available sequences in GenBank. Now I'd also add that I would cross-check my 23andMe data to make sure they said the same thing. And if not I'd investigate why.

2) Build myself a personal browser. Probably a UCSC Browser because that's what I know best, but I'd consider others. [well, ok, I'd hire one of you guys to do that, technically]. There's a bunch of things I'd want to consider to add as tracks and annotations, but I haven't thought them all through yet. I'd also plan to create custom tracks on my own reading/research that I want to link to regions/genes/variations. Like a literature track, I think. This would be how I'd store stuff I want to access later, and accumulate over time. Personal curation, I guess.

3) Look closer at well-characterized and medically-relevant genes probably based on the NHGRI catalog, GeneTests, etc. I know this is looking under the flashlight, but it would still be the part I'm most curious about--and the best chance for new understandings and actionable stuff. I'd add that I'd also look specifically for CNVs in my data too.

Your turn. What would you do?

EDIT: the question software wants me to offer a bounty now. Seriously--no one has thought about a workflow for this kind of data? I found this discussion fascinating, but I was really looking for the outlines of a process. If I can figure out the bounty thing maybe I'll add one....

EDIT 2: I started a bounty. Although the software told me it was going to be a 50, it says 100--whatever. I'm looking for a workflow--a series of steps you would perform to explore one person's whole genome data if it was given to you. It doesn't have to be yours. And it doesn't have to be written in Perl. An outline is fine.

[Gww]

Try to avoid freaking out when you find a bunch of markers associated with diseases?

[Mary]

Heh. Step 4: obtain counselling.

[Neilfws]

Should this question be community wiki?

[Mary]

I don't know. I thought it was a real bioinformatics process question, but as there may not be a single answer perhaps. I'll switch it.

Answer 1

[Istvan Albert]

I may be in minority but I'll say this: right now I simply don't want to know - Did you ever notice how genomic variation never correlates with good news. It seems there is only bad news. There are no SNPs for happiness, friendship or love.

I think the chances of reducing my own quality of life by supposedly finding something worrisome that later turns out to wrong/incorrect is far larger than the chances of improving the quality of my life by the same token.

[Michael]

+1 for this alone: "There are no SNPs for happiness, friendship or love." However I was thinking, maybe there are such SNPs, they are just not interesting for research.

[Mary]

Thanks for that perspective. I was very reluctant, but finally decided that there were too many conversations in the field that I was unable to participate in, so I picked up a 23andMe sale. But I'm not eager for my whole genome either, actually. But the question really was for professionals in this field--if you had it, what would you do?

[Chris Evelo]

I would probably do exactly what you did. Ask other professionals in the field what to do.

[Chris Miller]

Ignorance != bliss in my mind :)

[Istvan Albert]

Don't get me wrong if someone has some problems that they would like to better understand then personal genotyping is a great way to learn more. But for healthy person looking at this data - I think it can do more harm than good. This is born out of practical medical advice as well, I have heard credible opinions that overly proactive prostate cancer screening is actually more detrimental than beneficial (most people with the disease would die long from natural causes other than the cancer while the treatment greatly degrades their life) - and this could be true for many other diseases.

Answer 2

[Brad Chapman]

It helps to have a biological question in mind; something like: "my grandfather had Alzheimer's disease, what are my risk factors?" This is not any different than primary research work. It's difficult when a researcher says: "I did some microarray work/sequencing on this system; tell me what is interesting about it." Unless the answer is obvious or well known, it will be a struggle to separate the signal from the noise.

As a practical example, I had a 23andme genotyping done last year. I learned some useful things like a predisposition to Coeliac disease; this explained why not eating grains had such a positive impact on me. However, I'm not an especially interesting person so didn't have other avenues to dig deeply into.

In contrast, we also got genotypes from two of my friends with autoimmune diseases. I've been able to look into these more and discover some connections beyond what the 23andme analysis provides. This makes me feel good to be able to do something to help, and hopefully longer term may have a positive impact on their treatments.

This type of in-depth analysis will become more common as more people have sequences and there is demand (and an emotional appetite) for it. I think it will be the initial useful direction for personal genomics.

Finally, I believe open-science researchers can play a roll in allowing people to embrace their genome information. If we take control and be as open and proud of our genome as we are of other aspects of our personalities and lives, this can go a long way towards avoiding potential future exploitation. It's ultimately an individual decision, but the number of people willing to share their genomes so far is encouraging.

Answer 3

[Neilfws]

I recommend visiting 23andYou and taking a look at the 3rd party tools.

A couple that I tried and liked:

• Promethease: runs your SNPs through SNPedia (Windows only; I run it in a VM)
• Interpretome: does all kinds of general, clinical and ancestry analysis but keeps your data on your machine

[Michael]

"Make sure you don't miss any 23andMe emails by adding [e-mail] to your address book." LOL

[Mary]

David has already run his Promethease. According to his blog post he holds the current record for the most genotypes. There's also a quote over there from Mike Cariaso about it.

Interpretome, though, includes that retracted longevity paper data as one of the possible analyses. I don't know how that's being handled. Last time I ran it on my data there was no way for a user outside the field to know that.

[Mary]

Separate question though: do you think most of the tools are prepared for whole genome data?

[alex]

https://dnalifestylecoach.com[1] is another way to analyse your 23andme data

Answer 4

[Bioinfosm]

I would start with something like the analysis Complete Genomics provides. All your genome sequence data mapped to the reference sequence genome. The results would contain all the differences (SNV or indel) of your genome with the reference sequence.

• Annotate all those differences to genes/pathways
• be able to filter them for type of difference (nonsense or inter-genic or frameshift, etc)
• Copy Number Variation - to determine if your genome is missing some regions of the genome or has extra copies of some other genomic regions
• Structural variation - determine if there has been some shifting or relocation of your genomic material with respect to the 'normal' reference genome
• be able to link all these changes to known/published reports on phenotypic/medical effect of these changes
• have some kind of a visualizer (IGV from Broad is very useful) to gaze through the data, especially for the changes of interest
• as someone suggested, 3rd party tools of 23andme, or other resources with information for predisposition of a particular condition
• check with a genetic counsellor (not all predisposition is bad :)
• genome is just one aspect of you; to get a more complete picture, one needs transcriptome, methylome and microbiome (at the very least!)
• multiple resources are becoming available of variants seen in populations; so some of your variants may already be present in the 1000genomes project, dbSNP, 200 danish genomes from BGI, etc and are thus likely 'common' variations
• as someone suggested, make sure of the privacy of all this information; you never know!!

[Mary]

Thanks for that. Most of us haven't had access to Complete Genomics data yet (I presume), so we don't know what it has or doesn't offer to individuals.

[Bioinfosm]

Mary, they have a public release of their downloadable datasets = http://www.completegenomics.com/sequence-data/download-data/

BTW, I am in no way endorsing Complete Genomics. One can add a lot of annotation to the variants, but theirs is a really good single comprehensive analysis solution for whole genome data.

[Mary]

Sorry, I meant their data in analysis suite.

[Mary]

Ok, I'm gonna give you the bounty for the only one offering a multiple and diverse steps! Thanks for the info.

Answer 5

[Michael]

Whatever you do with it should be up to you to decide, use it for your personalized medicine if you wish. So my sole recommendation is: Keep the data private and well protected and encrypted! Decide in an informed way, whom you grant access to them.

If such data (one day) allow for serious diagnoses I wouldn't want my insurance, my current employer, new potential employers, my bank, etc. to be able to draw conclusions that might not be based on solid grounds at the moment. Data privacy is the most important concern, imho, as you cannot call the data back.

[Mary]

I'm not one of the people making my info public either. I keep winding up in arguments with people about this. Often I find the people I'm discussing with come from places with some form of national health insurance. The answer may vary by location.

[Istvan Albert]

This is a very good point - I think attempts to abuse this data will be pervasive.