Shared publicly  - 
A few initial thoughts on 23andMe's new $999 exome product:

1. The company steers clear of the FDA's baleful gaze by (at least initially) offering raw data only with no analysis.

2. This is for hardcore beta testers only. A raw exome sequence with no interpretation requires some serious data-wrangling to extract useful data from. Even existing open-access interpretation tools like Promethease will provide fairly overwhelming data.

3. The offer is open only to existing customers. That has two benefits: the one they emphasise on their site is that it selects for people with at least some experience in looking at genetic data. A more important (but unstated) benefit is that the customers' chip data will provide a comparison data-set for exome quality control.

4. We shouldn't forget that Knome has been offering both genome and exome sequencing DTC (along with interpretation) for a long time. However, my sense is that they've attracted relatively few customers. I suspect 23andMe will be able to rope in a substantially larger customer base even at a comparable price point, leveraging their marketing and community-building strengths.

5. Still, no question that the number of people willing to fork out $999 for a non-clinical-grade exome is pretty small.

6. At Genomes Unzipped we'll be looking to get hold of at least one of these exomes, as well as providing some suggestions on ways in which the raw data can be used. Exciting times for those of us who've spent the last few years working with sequence data...
Nils Homer's profile photoDaniel MacArthur's profile photoDeniz Kural's profile photoDan Frost's profile photo
I am glad I waited. The BGI $999 really made me think a few months ago. There's no motivation like analyzing your own data.
Awesome, check or CC?
Also, who wants to write down a brief guide how to analyze this using public databases?
+Adam Olszewski I think we're going to try to coordinate a guide over at Genomes Unzipped, if someone doesn't beat us to it...
Agree with your thoughts, Daniel. Few other comments/questions:

A. Not sure Knome is offering DTC anymore (can anyone confirm or deny?)

B. Does the difference in price ($999 for 23andMe exome, $9,500 for Illumina whole genome, and $7,500 if medically relevant) make sense? That is, do you get data that is 1/10 as valuable by paying 1/10 as much for an exome vs. a whole genome, or is saving up for a whole genome a better investment?

C. Will the provision of raw data DTC truly keep the FDA at bay? They've indicated in the past that they aren't interested in going after raw data - and they have plenty of more pressing targets to consider, including 23andMe's standard service - but 23andMe is certainly pressing the issue a bit.
Hard work needed.

I would like to do compare exomes of all family members - when it gets cheaper.

If/when enough people do this do 23andme retain the right to analyse all the data, compare exomes, discover new SNPs and rare variants, determine gene haplotypes, etc?? that could be a useful - in the old days (10 yrs ago) companies like Genaissance spent their $150 million IPO money on exome sequencing just 100 genomes! Got to approx 8,000 genes before having to stop.
I just asked myself the same question +Daniel Vorhaus did (B): Why would I want to pay ~$1k now for an exome, if I could get so "much more data" for ~10x the cost from an Illumina complete genome (or could just wait for another year for prices to drop even more ;)?
I'm hoping to be among the "chosen". If so, I will post the results as I'm among PGP1K
+Bastian Greshake do both your exome and whole genome. Do it many times. Validate validate validate. Also take a look at the exome capture comparison paper that showed that some important mutations were missed by WGS but picked up by exomes. I would be happy to help with a communty pipeline to look at peoples data.
Massive +1 to +Nils Homer's comment. Don't believe anything you get from sequence data until it's been independently validated. It's not that the technologies are crap, but the genome is a big place, so even a low error rate will give you plenty of things to worry about.

Also bear in mind that the false positive rate is even higher for functional variants than for the rest of the genome (since evolutionary constraint means the prior on a functional variant being real is lower than the genome-wide average).

Nils, we'll be in touch regarding a possible pipeline...
+Vladimir Gainullin Good question :) +Daniel MacArthur +Nils Homer May I join the community pipeline? I'm working on a system to host & share such things semantically. I made a pipeline with BWA+GATK for now hosted online for exomes, but it's pretty basic.
Hey Daniel - I know this is an old post, but thought I would update you on something we have been working on.  Our head of R&D got his trio sequenced and wrote a cool post about his initial findings (it helps he and his team created a sweet new viz tool for looking at this stuff).
Add a comment...