I’ve decided to start my own personal journal club. This is the first meeting. Basically, I’m -a- read a science article once in a while and then try to summarize why it matters.
I’ll hide this sort of stuff under a cut – in case there are any science allergies among my readership.
This is one of two recent papers in which clinical researchers did complete genome sequences of individuals in pursuit of clinical goals. I.e: This was no longer genome sequencing for its own sake – in the last five years whole genome sequencing has become a usable tool in the arsenal of the geneticist.
Charcot–Marie–Tooth disease is a neurological disease that appears (to this interested layman) to present in a baffling array of symptoms. I see wrists and ankles, flexion and extension, excessive tension, lack of strength, and so on. It also appears to be genetically complex – with various pieces appearing dominant, recessive, carrier, and all the other terms that I vaguely recall from my genetics class.
In other words, a perfect candidate for a whole genome analysis.
They used a particular analysis platform, the SOLiD sequencer from Applied Biosystems. This is one of three major players in the 2nd generation sequencer market – and the one that according to the grapevine (manufacturers are very cagey about how many instruments they have actually sold) has near total market dominance. The authors allude to the fact that it’s difficult to get a reasonable estimate of the cost to repeat this experiment – since cost is constantly dropping. When they wrote, it was perhaps $50k for a decent coverage on all 3 billion base pairs. For the record, that’s perhaps two “runs” of the instrument – at about a week per run. They got 30x coverage for their money. The human genome project thought they were doing quite well to get 10x. Threefold better coverage, in weeks, at a fraction of the cost. Now that’s progress.
Again, this has become a not unreasonable scientific experiment. It’s not quite clinical yet – but certainly clinical research.
They augmented the raw base pair reads with two different sorts of microarray – and used some pretty old-school methods to do the analysis. ClustalW and BLAST were old when I started learning this stuff. This speaks to something that computer / algorithm types don’t like to hear: At root – this stuff is about the biology. The important algorithms right where the rubber meets the road in biology are the same today as they were back in 1985. The chemistry has improved, and the cost of doing the data collection has dropped. There are new algorithms needed to deal with overwhelming amounts of data – but for cases of a few sequences – BLAST and ClustalW still work just fine.
Out of millions of point mutations, they found around 174 that seemed to matter. They then went back (I think – I’m a little fuzzy here) and got genetic data at those points for the rest of the subject’s family. They used that information to identify 50+ points of genetic variation in two proteins that appear to explain the pattern of inheritance seen in this family.
They point out that a classical genetic screening for this disease can cost more than $15k. This means that within a year or two, it will be cheaper to do a whole genome sequence than to do a standard screening.
This is the first paper where, when I heard about it I said “oh my gosh, in my lifetime.” Fascinating.