BioInformatics: IBM Resources and Research

I am very excited about Dave Tribbett' devotion to a series of deep analysis of postmodern genome informatics - as a core of critical mass, producing an explosion in the life sciences. I believe Dave’s present article is also great. Formally, there are only minor errors (for instance, to correct that Sun Microsystems is "now part of IBM" - actually on January 27, 2010 Sun was acquired by Oracle Corporation for US$7.4 billion, based on an agreement signed on April 20, 2009).

I more than agree to the "big picture" that "Life Science computing" is for IBM to lose - just as the PC was an "IBM PC"; just to be lost to Microsoft - that was a very young and aggressive company compared to perhaps too big and at time ossified IBM. I have written years ago a prediction of the "Big One"; and earthquake-like effect whenever the tectonic plates of "Big IT" pile up also private "Big Pharma" upon "Government Genomics". That predicted time is now.

Indeed, with great devotion and expertise, Caroline Kovac (recently retired) built a spectacular "IBM Life Science" program for at least a quarter of a Century - and as Dave’s embedded YouTube overviews shows, a huge array of activities in the Life Science Computing Program is thriving at IBM. I am particularly familiar with the trailblazing work of Isidore Rigoutsos and his colleagues; aiming at novel pattern recognition algorithms to address DNA structure.

Yet, a daresay, that the very enormity of the wide-spectrum Life Science effort of IBM might similarly result in IBM losing this game, the winner of which (just like in the early days of the "PC") just could not be predicted. In retrospect, a Monday Morning Quarterback "wisdom" might come to the crystal-clear "analysis" that the monstrous (and monstrously complex) IBM lost a singular focus. They wanted to accomplish holding both the key to the hardware architecture as well as (belatedly) gaining the upper hand of the PC OS (famed OS/2). They ended up with securing neither - and long time ago sold the entire IBM PC division for a token price. Likewise, what is the Government's slice in the pie of "PC"?  - Just about zero.

Now take Genome Informatics. I was thrilled to hear about the IBM "Blue Gene" at the Monterey 50th Anniversary of the Double Helix (2003 February, exactly 7 years ago), from Caroline Kovac. Observing my surge of enthusiasm, she felt it necessary to utter some words of wisdom as a "caveat". She reminded me that the "Life Science Division" she was leading was NOT in control of Blue Gene at all - the World's fastest supercomputer (at that time) belonged to the IBM' "Computer Development Division"... Those of us who ever worked in huge operations like IBM (or the US Government), know that divisions of an entity are compelled AGAINST their cooperation, since in fact they pitch their efforts to COMPETE AGAINST EACH OTHER for one single pool of resources of the entity.

Thus, I am less than fully convinced that any single existing "Big Whatever" (including US Government...) is going to win "Genome Informatics". If history is any lesson, the winning combination is more likely to be laser-beam focused EMERGING entity (like Microsoft was, at that time) that grabs the singularly most crucial tenet (that was the OS and killer apps software for PC) and develops global alliances with just about all participating entities (minus perhaps IBM and Apple that decided NOT to take part in the cooperation). Far East manufacturers wisely opted in – and came out as winners of hardware production business, also Intel decided to focus on pure-play of CPU serial chip design and likewise became a big winner.

So what is the focus of "Genome Informatics"? Chances are that one answers: "DNA sequencing". Wrong! Affordably revealing the full human genome is a necessary, but not sufficient step. To take the cliche, "Genome Projects" are often compared to the "Moon Shot" (see e.g. Nobelist Sydney Brenner's very recent essay), in which comparison he says is quite literally true: "Getting a man on the Moon is relatively easy". "Getting the man safely back from the Moon is the harder part".

The AAAS grand annual meeting is wrapping up today - and Silicon Valley is teaming with not one but two most successful DNA sequencing Centers (Complete Genomics and Pacific Biosciences) rolling affordable full DNA sequences from the assembly line, like Ford T models avalanched the US from Detroit. (It had to be matched, for sustainability with an entire network of gas-stations…)

While I was not attending AAAS in San Diego, in the coverage I did not detect any major speech pinpointing that DNA sequencing does not reveal the "Language of Genome" at all. It does brings into plain view all the A,C,T,G “letters”, like when you obtain a copy of "War and Peace" with all Cyrillic letters of the Russian language on display - yet “readers” (except those who mastered the Russian language) will understand absolutely nothing of what the overwhelming number of letters might mean.

There are plenty of "Genome Sequencing Centers". To win "Genome Informatics" we need one more "Sequencing Center" as a hole in the head. Instead, we desperately need a private domain core-company that is totally focused in Genome ANALYSIS, such that a "Center of Genome Interpretation" emerges around it, reaching out to all R&D and business of the land.

Absolutely (just like Microsoft's OS and Killer apps), the core of this crucial fulcrum will not remain isolated, but will spread to virtually all hardware and software companies that are eager to grab a slice of the PostModern Genome Informatics market (as big as global Genome Based Economy is...).

Take the mentioned Intel. They bought into Genomics (now "Informatics") by contributing to $100 M investment to Genome Sequencing by PacBio (2007), plus put together a (small) "Downstream Data Analysis Group for Genomics" - that Intel disbanded when the global financial crisis hit us hard. The rationale might have been that "let's focus on sequencing first - the downstream analysis can wait; besides, as long as anybody does it with boxes with ‘Intel inside’ on them, the IntelVC investment in sequencing is safe". Moreover, technically it would take Intel just a few days, weeks or months (at most) to re-assemble an order of magnitude more potent "Downstream Genome Data Analysis Group". However, de facto, organizations of  huge industries hardly ever move that fast, since decision-making is hierarchical and needs elaboration, submission and evaluation by committees of umpteen layers of corporate structure. Thus, it remains an open question if the huge freighter ship of IBM or Intel might turn tighter corners.

Meanwhile, Microsoft, Google, HP, Dell, Oracle could "cut in" - since all of them put together their Life Science programs, for readiness for "health-care business" - for the uncertain time when the Government might be ready with reforms accelerating e.g. digital health-data repositories. It might be debatable, though, if any of them singularly focus on "DNA Functional Analysis" - as it takes a quite unprecedented multiple domain expertise of disparate fields, that need a "psychological welding" (borrowing Nola Masterson's term) to get the two hemispheres of "genome informatics" together (my favorite term refers to the much drier anatomical structure of "corpus callosum"; the massive bundle of cable system connecting the right and left hemispheres of the brain).
So, where is the Government? Any Program at any of the countless branches of the organization where "Algorithmic (software enabling) approaches to interpretation of genome function" are elevated into the role this issue ultimately will be?

Hardly.

The Government is not geared to take the enormous risk that scientist might THINK (as some, indeed, might not). The Government's role is to program of what contractors DO. But it is all right, since actually some "Genome Computing Architectures" (glimpsed e.g. in the first seconds of http://www.youtube.com/watch?v=mSRMCDCVg6Y ), a "Nurture server" (ingredients of what UPC barcoded nutrients contain) and a "Nature server" (knowledge-base of dietary consequences of genomic conditions and environments) are not only eminently doable "on the cloud" - but both USDA and OSHA are actually well under way of having implemented (part of) it. Since ALL government R&D (in the USA) is paid from our precious tax-dollars, it is just a simple mandate to issue that any/all government-funded research results pertaining to both "Nurture" and "Nature" aspects of genomics must be uploaded to a government-maintained cloud computing - and made available for US individuals and industries (sorry, not for Countries that don't pay taxes to support such repository of strategic value).

The author of this blog-reply, having decades of experience in Academia, Industry and Government (see bio at http://www.usa-siliconvalley.com) realizes full well that the above might be perceived (and even mis-labeled) as a "pipe-dream" like "the Manhattan Project" could be (wrongly) labeled as the "pipe-dream" of Albert, when he signed Leo's letter to the President Eisenhower on August 2nd, 1939. Yes, the "Plan" received an  initial pittance of $4,000 from the Government to get going - but ultimately changed history.

Pellionisz_at_JunkDNA.com