23andMe has the most will to succeed, followed by Navigenics, followed by deCODEme. All three have sufficient potential funding, so will (and luck) will most decide who will survive The Chasm.

Yesterday, I mentioned a popular business graph called “The Chasm.” The Chasm is start-up business jargon for the difficultly businesses tend to experience growing from a market of early adopters to the general public. This is because customer motivation changes: early adopters buy because they like new technology, but most people buy because they want to solve problems with minimal effort.

Today, DTC (direct to consumer) genomics is still in its “innovators” market phase, though continued coverage in Wired and regulatory attention suggests that the market is approaching an “early adopter” transition. But which genomics start-ups will survive to cross The Chasm to reap the riches of a greater market?

Two factors keep start-ups alive during tough times:

1. Funding: No money, no payroll, no people. Start-ups, particularly venture-funded start-ups, are profitable until they are bigger, more mature companies.
2. Will: How much does a start-up want to succeed, and what do will its leaders lose if it doest? Does your company have the morale and reputation to recruit talent and investment to beat the competition, weather setbacks, and persist through regulatory struggles?

Consider the “big three” DTC start-ups: 23andMe, deCODEme, and Navigenics. Other competitors are possible, but identifying them is speculation. Further, the recent California “legal lab” crackdown seems to have scared away most other scrappier competitors for now.

I think that all three competitors have ample funding… if they have the will to spend it. I argue that 23andMe and Navigenics have that will, while deCODEme may or may not.

The leaders of 23andMe and Navigenics are most personally and publicly invested in the success of their ventures and thus are most likely to succeed. 23andMe wins the accountability metric because if it doesn’t succeed, it will forever be known as “that Google’s wife’s start-up toy with that disgruntled affy chick.” These women probably do not appreciate being known as such, and are powerful and determined enough to prove otherwise. That alone will keep 23andMe around indefinitely. The rest of the 23andMe team is also well featured on the about page. However, the iStockPhoto slideshow on the 23andMe team page needs replacing.

Navigenics team is also very well featured, even better than 23andMe’s team.

At deCODEme, Kári Stefánsson may publicly represent the business, but he’s the CEO of deCODE. Who is personally accountable for the success of deCODEme itself? On both the old and the new versions of the deCODEme website, nobody is named. The new About deCODEme page does feature a photo of the deCODEme team, but the only names are of deCODE researchers publishing papers, not deCODEme management. (they are pretty nice photos, though)

Further, both 23andMe and Navigenics feature recruitment on their websites and actively advertise positions with third parties (a quick Google search confirms this). deCODEme does not.

Finally, deCODEme’s parent company, deCODE, has not been doing well financially and has never reported a profit. It has recently eliminated many positions, is debt-leveraged, has sold-and-leased its American office, it’s stock price is at $1 and cents from about $28 in 2000, and its CEO has warned of ending operations. All of this is bad for morale, and if more cuts must be made, an unprofitable deCODEme is a likely candidate. I doubt deCODEme will ever be eliminated because it’s obviously Kári’s personal initiative, and as far as I can tell, deCODE is Kári. What’s most likely, if things get bad, is that deCODEme will process orders, but languish without growth or direction as 23andMe, Navigenics, and other competitors continue to hire, grow, and improve.

… Continue Reading »

deCODEme Genome Browser (demo account)

What if you could see your genome DNA like sound tracks in mixing software? What if you wanted to listen to music, but could only play it in mixing software? With deCODEme’s new genome browser, now officially launched, you can.

Genome Browser demo video

I previewed the Genome Browser in Reykjavík a few months ago, and my response was then as it is now: “No.” Now, I like deCODE, I love this industry, and I want people to be excited. So, I get a bit overheated when I see something with so much potential grow in the wrong direction, like some geeky math genius with basketball dreams. Put that bling where it belongs, son: in the bank, not in your teeth.

and… that’s deCODEme’s Genome Browser: right bling, wrong places.

The Bling:

Genome Browser is the best, most accessible, most personally relevant genome browsing software today. I can seamlessly zoom from chromosome, to gene, to individual nucleotide sequence. I can see where deCODEme SNPs are geographically aligned in the genome. I can search for a term, like “CCR5,” and Genome Browser zooms to results in real-time. And, I can cross-reference against the top three genomic public databases. Very cool.

Similar Products. Note that these are scary, ugly science tools:

• UCSC Genome Browser
  
• NCBI Reference SNP
• Ensembl SNPView
• NCBI Sequence Viewer
  
• NCBI Map Viewer

The Teeth:

So who are the users of Genome Browser?

Scientists? I can’t reference it. I can’t link to it (because of Java). It doesn’t have the technical depth to be a better scientific tool. And, I can’t import or export tracks —my own data.

Amusing, but not useful.

A tool for casual enthusiasts? I have to wait several minutes to install Java and wait fifteen seconds to load the application? (waiting is bad) I have to read instructions because the interface isn’t self-descriptive and intuitive? (reading is bad) Where are the hyperlinks? Where are the scrollbars? Why can I do something annoying, like drag rows to arbitrary positions, but I can’t do something useful, like click-drag zoom directly in tracks? Why is there a mostly empty row for every deCODEme phenotype instead of a compact, single row?

Confusing, and not useful.

Ultimately, deCODEme’s genome browser is like an average modern linux application: written for its programmers’ gratification, not for its users’. That’s something not even the shiniest transparency effects can fix.

The Bank

Genome Browser has value, but it’s misplaced in a bad interface. This can be fixed:

1. Replace the Java application with AJAX and/or Flash and make Genome Browser work like a regular web app. UI Design is not a common scitech industry expertise, but one can do well just by avoiding mistakes. Learn from Google: keep it simple, fast, and useful. All the brushed-gray gradients in the world won’t make you Apple.
2. Fix the track system for phenotypes to work in a single row. Avoid vertical scrolling.
3. Label all terms with tooltips and hyperlinks. Avoid noisy jargon. For example, replace “known Genes from UCSC” with “Genes.” (details like this belong in documentation)
4. Keep symbols consistent. For example, SNPs and sequences look the same, but SNPs are diploid threads, sequences are haploid pairs.
5. Seriously consider an “Import SNPs,” plugin, or API feature.
6. Buy some pizza, set up tables and laptops in the university yard, and watch students use Genome Browser and deCODEme. It will be good PR, and you will learn what users want.

Conclusion

Eventually, hopefully, someone at deCODEme, 23andMe, or Navigenics is going to realize that their product is a crude, disposable prototype for cheap full genome sequencing, and that their actual job is market finding, not product building. Thus, personal genome products TODAY aren’t the REAL business, they’re just early loss-leaders for building company brand and expertise.

So, if I were the CEO of deCODEme (or 23andMe), I wouldn’t budget a penny for any deCODEme project, including new releases for the deCODEme Genome Browser, unless my team could tell me in three sentences how that project would:

• Help discover and develop the market for consumer genomics
• Be relevant to the company in five to ten years
• Give the company an advantage over its competitors once the market is understood

I am having a difficult time composing those three sentences for this release of deCODEme Genome Browser as it is now.

MY SNPedia REPORT

Services like 23andMe and deCODEme test for hundreds of thousands of SNPs (genetic datapoints). Yet, these services only attempt to interpret less than 100 traits in their web reports. [1]

Fortunately, these services allow you to download the raw data from your test to be interpreted elsewhere.

For example, recently Mike Cariaso “creatively profiled” Lilly Mendel (23andMe’s female demo user) using the public SNPedia database and Promethease software to win the 23andMe Win Your Genome Contest.

• Lilly Mendel’s Promethease Report
• SNPedia (wikipedia-like public database for the human genome)
  
• Promethease Home

These tools are not yet user friendly, and they certainly exist beyond the general consumer. However, I purchased a 23andMe test a couple months ago, and as a power user, I wanted ALL the data. Yes, I know that the research is new, the interpretations can be highly speculative, and 23andMe is making a laudable effort to make the best research accessible. But hey, I spent $1000; I want to know everything. Further, while these services do not report many high-penetrance genes like BRCA (breast cancer) for legal reasons [2], some of that information is in the raw test data “for your own interpretation.”

If 23andMe is the Apple I of genomics, this is the box of parts for garage assembly. Enjoy!

How To Use SNPedia for 23andMe and deCODEme

1) Get Data

Download your test results. If you do not have a 23andMe or deCODEme account, you may download a sample dataset from 23andMe’s trial users. (deCODEme’s trial user does not yet support data download)

Download Lilly Mendel’s raw data from ThinkGene

23andMe:

1. Login (either as your account or as a Mendel trial user)
2. Go to “Browse Raw Data” in the left menu
3. Click “download raw data” link at the top right of the “browse raw data” page
4. Enter password, secret question, select profile, and click “Download Data”
5. The download will be about a 2MB zipped text file.

Here is what raw 23andMe data looks like from my genome:

# rsid       chromosome    position    genotype
rs3094315    1             742429      AA
rs12562034   1             758311      GG
rs3934834    1             995669      CC
rs9442372    1             1008567     AA
rs3737728    1             1011278     AG
rs11260588   1             1011521     GG
rs6687776    1             1020428     CC
rs9651273    1             1021403     AG
rs4970405    1             1038818     AA
rs12726255   1             1039813     AA
rs11807848   1             1051029     CT
rs9442373    1             1052501     CC
...

and so on for over 570,000 lines.

Raw deCODEme data looks like this:

Name,Variation,Chromosome,Position,Strand,YourCode
rs2278544,A/G,2,136262580,+,AG
rs4954633,C/T,2,136263105,+,CC
rs3213890,A/G,2,136268658,+,AG
rs1807356,A/G,2,136271954,-,AA
rs2015532,A/C,2,136271995,-,AC
rs2322659,C/T,2,136272129,+,CT
rs3769013,A/G,2,136272652,-,AG
rs2304371,C/T,2,136278027,-,CT
rs2304370,C/T,2,136278205,-,CT
rs3739022,C/T,2,136278942,-,CC
rs12988076,A/C,2,136286318,+,AC
rs6719488,G/T,2,136291669,+,GT
...

and so on for over 1,000,000 lines.

2) Get Software

Download Promethease (windows only, version 0.1.14) from SNPepida. Also, see Promethease’s homepage.

3) Run Software

Launch Promethease and select your data file. Select your race. Begin Analysis. Note that Promethease automatically detects which service generated your SNP data.

WARNING: generating a report with Promethease may take hours. If there is demand (and time), I’ll write / help write better Open SNP software. Please contact me if this interests you.

4) View Report

MY REPORT: Andrew Yates’s SNPedia report via Promethease [3]

Promethease outputs a final report of about SNP 1300 entries as an HTML file to your desktop. It’s not pretty or easy to read —let alone understand as actionable medical information. But this is the best that Open SNP software has to offer today.

See Interpreting Promethease / SNPedia Results for help understanding your Promethease report.

5) Report Reflection

While my SNPedia report contained more information about my SNPs, I didn’t learn anything worth knowing that 23andMe (or deCODEme) didn’t already tell me. Further, running and understanding my “open SNP” report, while interesting, took hours. While much potential exists for community efforts to understand our genome, SNPedia is not yet ready for the majority of genetics enthusiasts, let alone the general public.

Notes

[1] 23andMe and deCODEme officially report less than one hundred traits, plus ancestry. 23andMe tests for about 500,000 SNPs; deCODE tests for about 1,000,000. However, the useful information in both tests is about the same (for now).

[2] The industry line is that users shouldn’t know about some high-penetrance genes for reasons including “it’s for your own good” and “high penetrance genes are not common enough,” but the real reason is that these genes are defended by walls of patents, expensive licenses, and litigious companies. See “Low penetrance genes v high penetrance genes”

[3] At first, I was leery about posting my genome online, but then I realized: hey, if I’m ever discriminated against (and I can easily check my server logs to see who has viewed my genome report), maybe I’ll get to participate in a landmark judicial case!