Monday, January 14, 2008

Promethease deCodes 23andMe

Promethease annotates your genome using SNPedia.

It is possible thanks to the Personal Genome Explorer which allows 23andMe customers to export all of their 550,000 genotypes. Promethease reads this export format, and uses it to generate sample reports.

Your genotypes will probably be a bit more boring, since these random files get some pretty weird diseases. The report format isn't yet as intuitive as it should be, but you'll manage.

Your changes to SNPedia will appear in future reports. To generate a report simply drop an export file onto promethease.exe . From the DOS command line you can compare to other hapmap populations, instead of the default CEU. Runtime is about 2 hours. It doesn't eat much of your net or cpu, it just moves slowly to be courteous to others.

PGE does not yet provide information to orient snps relative to dbsnp. As a result many of the genotypes are flipped. This causes Promethease to ignore population frequencies. These SNPs will show with a 'None' along their left side.


Tim Read said...

Looks interesting. Must get my own genome SNP-typed sometime.

Unknown said...

deCODEme and 23andMe have now both begun to release customer data.

Promethease is now at version 0.1.2 with the ability to autodetect and support these 4 formats (PGE, Promethease, deCODEme, and 23andMe).

SNPedia's page on Promethease is now available.

Unknown said...

Thanks for the pointer to this. A couple questions... when I run it, I get a lot of 'keyerror on magnitude' lines. Bug?

Also, can you explain the output web page a little? What does the number (or None) to the left of the rsid represent? And, to the right of the rsid, only some of the SNPs have risk factors. Is this because the data is incomplete or inconclusive?

Simon Lin, MD said...

Hi Michael,

I like the word you coined "23etAl"! So I commented it in my blog below.


Unknown said...


I suspect you were running 0.1.4. Promethease is now at 0.1.5. It's reports link to this results explanation

'keyerror on magnitude' is a clumsy note that it looked for, but did not find a particular piece of information. In time this and many other warnings will disappear, but for now they provide useful bits for debugging.

The numbers at left indicate HapMap frequency for this genotype. Lower numbers indicate rare genotypes. As you've guessed, missing data is missing at least from SNPedia, but often the science itself is not yet firm enough to convince anyone.

John said...

How often is promethease updated? What databases does it use. Is there a webpage that explains this? How dependable is this database?Can it be used for peer reviewed publications? Pls comment.

Unknown said...

How often is promethease updated?

Currently there is no release schedule. When I feel like I've got something worth releasing I release. This could be 3 in a week, or nothing for 2 months. If your version is more than 30 days old its worth looking at the Promethease page to see if there is a new version. Since the SNPedia database is constantly being updated, and your report will reflect the contents of SNPedia at that moment. A future version of the app will be better about alerting you when a new release is available.

What databases does it use?
Promethease uses SNPedia as its database.

Is there a webpage that explains this?
I'd like to say that you're reading it. But that's only half true. You're also writing it. If you ask a question here, or in the wiki we will try to answer it.

How dependable is this database?
Over the last 12 months dependability scores vary from 8.8 to 9.3 mu.

Can it be used for peer reviewed publications?

These researchers at the CDC seem to think so. If you choose to do so, you may wish to link to versioned 'Permanent link' found in the lower right corner of every SNPedia wiki page. This wikipedia article should further inform your consideration.

John said...

Cariaso, thanks for clearing my doubts. I was asking those questions as many people told me that SNPedia is not a database like dbSNP from NCBI, but its a blog and we may not depend on it. I have no personal doubt of my own. I am just a learning student from Mississippi. Again, thanks for sharing the info.

Unknown said...

Good of you to ask. I assure you many more lurk quietly, but wonder the same things.

They are correct that it is not like dbSNP, but it is also much more than a blog. Much of the information in SNPedia comes from dbSNP. But though a mix of human editors and software tools SNPedia is able to incorporate and cross check many other sources.

Your skepticism is a good thing, and I would advise you to continue and use it as you read the information in SNPedia. Please scrutinize everything, and if anything looks odd we all want to hear about it and get it corrected.

John said...

Cariaso, thanks again. I was wondering what Navigenics and decodeme are using to make those conclusions. I asked them, but they never gave me a straight answer. I was hoping that u might know something about it. DO u?

Unknown said...

They rely on many of the the same published research studies which you see in SNPedia. I am not able to speak about what additional considerations they might use.