Big Blue continues to weather industry storms and come out on top.
By John Russell
Sept. 5, 2008 | Trying to describe in a single story the full scope of IBM’s Life Sciences and Healthcare activities and related work being done by IBM Research is at best ambitious and at worst foolhardy. Big Blue may be unique in the corporate world in terms of the sheer number, diversity, and depth of its activities.
IBM revenues will top $100 billion this year, with services accounting for more than half ($54 billion in ’07). Although the Americas still make up the largest piece of the pie, Asia/Pacific, Europe, Middle East, and Africa collectively contribute almost 60% of revenues. Headcount is around 388,000, with rapidly growing encampments in India (94,000 by year’s end) and China (already IBM’s global procurement headquarters). By most measures—profits, cash, earnings per share, margins, etc.—IBM is in excellent health despite the macro economic malaise in Europe and North America and volatility within the IT sector.
Likewise, IBM Research is in fine fettle. The Thomas J. Watson Research Center in Yorktown Heights, NY, remains the impressive centerpiece, but is joined by seven facilities around the world in Zurich, Beijing, Almaden (CA), Austin (TX), Haifa, Delhi, and Tokyo. The company continues to spew out patents, 35,000-plus and growing.
Of course Blue Gene—IBM’s line of supercomputers named for grand challenges in biology—dominates the top 500 lists. The number of IBM scientists is modest, 3200, but extensive collaboration with scientists, academics, governments, and other corporations worldwide substantially enlarges their footprint and impact. Here are just a few ongoing IBM Research projects relevant to the health care and biomedical research:
STEM—The Spatiotemporal Epidemiological Modeler is a sophisticated tool to help medical and government executives identify, track, and potentially predict the course of infectious disease outbreaks. It’s now administered by the open source ECLIPSE organization. The underlying technology will likely find a variety of uses.
Checkmate—Together with Scripps Institute, Checkmate aims to better understand and simulate viral mutations—such as bird flu—to develop effective control and containment measures. Blue Gene-scale computing, new algorithms, and advanced microfluidics are all associated with this project.
Pyknons (Greek adjective for dense)—IBM computational biologist Isidore Rigoutsos is uncovering new functions for “junk DNA.” He uses term Pyknons for the vast number of non-coding DNA motifs identified by his pattern recognition techniques, which may have critical regulatory roles.
Blue Brain—Using the computational capacity of IBM’s eServer Blue Gene, researchers from IBM and Ecole Polytechnique Fédérale (Switzerland) are creating a detailed circuit model of the neocortex—the largest and most complex part of the brain.
The Genographic Project—IBM is contributing funding, science, and computational support to study ancestry of indigenous populations and volunteers (see, “International Blue Genes”).
This is only the glistening tip of Big Blue’s research berg. A vast array of other projects, virtually all of them undertaken with paying partners, hum away below the surface, tackling important life sciences questions whose solutions also present concrete commercial opportunities for IBM.
Yet today’s picture of cooperate robustness and research strength was much in doubt 15 years ago and is a testament to IBM’s ruthless and necessary re-invention. In the early 90’s, with large amounts of red ink flowing, many thought IBM’s best days were behind it. Sometimes history can be instructive.
Getting it Right
International Blue Genes
Several years ago, IBM raised eyebrows by partnering with the National Geographic Society to launch the Genographic Project (see, “Taking a World Genography Test,” Bio•IT World, June 2005). Scientists scoured the globe to collect DNA samples from indigenous populations. Meanwhile, the general public could (and still can) purchase $99 DNA kits, allowing them to submit a cheek swab and learn about their family’s ancestry via the genotyping of the (paternally inherited) Y chromosome, or the maternally transmitted mitochondrial DNA.
Ajay Royyuru, research scientist and senior manager, Computational Biology Center, leads IBM’s contribution to the first published data from the Genographic Project. Two such examples were published in the American Journal of Human Genetics last spring.
In one study, IBM and the Genographic Project team collaborated with Lebanese American University researchers in Beirut to genotype Y chromosomes from more than 900 Lebanese volunteers. Each individual supplemented their genetic data with personal information including their self-described ethnicity and religion. “It is not seen as an intrusive question in that part of the world,” Royyuru says wryly.
The researchers compared the Y genotypes of the three major ethnic groups in Beirut—Christians, Druze and Muslims. One of the main conclusions was that religious affiliation has a great effect on patterns of genetic variation in the three ethnic populations in Lebanon than say geographic differences. But Royyuru notes that there is not “one genotype that is exclusive for any religious affiliation. All of these religious labels are from the last few thousand years. The genetic labels are tens of thousands of years old.”
But the genetic analysis provides compelling evidence of historical migrations to Lebanon. About 250,000 men are estimated to have reached Lebanon from Europe during the Crusades. A Y-chromosome haplotype called WES1 (a subtype of the most common western European haplotype R1b) provides strong evidence for the Christian migration during the Crusades. WES1 is found in 0.2% of European populations east of Hungary, yet accounts for 2% of the R1b haplotypes in Lebanese Christians. And yet it is completely absent in a database of more than 1,000 middle-eastern Y chromosome haplotypes. A similar trend is seen for another haplotype spreading from the Arabian peninsula into the Lebanese Muslim population.
A second study looked at mitochondrial DNA diversity, and concludes that the African Khoisan population diverged from the rest of the human mtDNA pool 90-150,000 years ago. Additional lineages merged with the Khoisan in the late Stone Age, about 40,000 years ago.
Royyuru says the Genographic project continues to evolve, but the focus remains on understanding deep ancestry. Proceeds from the DNA kits—more than 250,000 sold so far— are helping about a dozen legacy projects providing grants for indigenous populations. -- Kevin Davies
Michael Hehenberger, global solutions executive for life sciences/pharma, joined IBM in 1985. “In 1989, I became the worldwide opportunity manager for chemical/pharmaceutical R&D out of Paris, because this was an industry where Europe was very strong. I have to admit we failed in creating a sustained IBM business, but I also think we approached the challenge in the wrong way: I was given a lot of resources, but I was also given a constraint. You have to do everything to run R&D solutions on the mainframe.” Hehenberger believes IBM is getting it right this time.
The wrenching changes of the early ’90s initiated a transformation that continues inside IBM. Over time, the “product” mentality has been supplanted by a solutions and service mantra that emphasizes higher-value offerings and led to the jettisoning of its PC business. Its go-it-alone approach has morphed into a can’t-do-it-alone realization (60 acquisitions in five years) that is far more accommodating to partners and embracing of open source. Customers also stressed they didn’t care about technology (or brand) per se, but worried extensively about solving specific problems, efficiently and less expensively.
“We’ve learned a lot since 1993 through our own experience,” says Michael Svinte, VP global pharma/life sciences and an IBM veteran. “We’ve lived through a lot of complexity and change and transformation. Our clients in pharma five or ten years ago didn’t want to hear about that. It didn’t mean anything to them because it was a pretty good business, the Halcyon days of pharma. Now their challenge is different and our experience is a huge advantage in having those conversations.”
IBM weathered the storm and remade itself. While it isn’t a corporate or workplace nirvana, it is again vibrant, prosperous, and tackling eye-popping projects that few rivals could contemplate.
Indeed, now that health care and life science has emerged as a major opportunity, IBM executives feel they can argue that Big Blue has something constructive to say—not just sell—to large biopharma companies whose plight (patent expiry, soaring R&D costs) resembles IBM’s waning Big Iron days. Major consulting acquisitions such as Healthlink and PricewaterhouseCoopers (consulting) have bolstered that argument by adding necessary industry expertise (plus customers) to match IBM’s extensive technology capability.
So what is IBM up to these days in the life sciences and health care communities—a world whose boundaries IBM thinks are blurry at best—and how does its premiere research organization fit into this picture?
“We have a lot of resources,” says Svinte. “Our challenge is to select areas that we really want to go after. We have done that in the life sciences space.” Svinte cites IBM’s biennial global CEO survey as an invaluable resource in identifying key areas (see, “IBM CEO Survey Shows Hunger for Change in Life Sciences”).
Excluding for a moment IBM Research exotica—and even exotica must carry its own water these days—IBM has divided the health care and life sciences space into three targets: Life Sciences/Pharma; Provider (health care delivery); and Health Plan/Payer. Within each bucket, IBM sets a “strategy focus” from which flows productized solutions.
For example, the strategy in Life Sciences/Pharma is to attack in three areas—R&D productivity, regulatory compliance, and operational efficiency—and deliver solutions aligned for each strategy bullet. A good example is IBM’s SCORE—Solution for Compliance in a Regulated Environment.
Document management is certainly a major part of it, with all the bells and whistles you’d expect from IBM and which are required by FDA. But powerful image management capability is also part of SCORE as is “track & track” functionality which can be used to monitor and secure supply chains. A recurring theme is co-development of these solutions with major clients. In fact, IBM has a SCORE-Bio-Medical Imaging Solution that fell out of a collaboration with Merck.
“In one big clinical trial, where imaging was a major component, just the [Merck] budget for mailing and postage and handling was more than $5 million,” says Hehenberger. “We can manage all kinds of documents. We also had a solution coming out of IBM Research where we had a way of managing bio-medical imaging data. So we combined that with our SCORE solution and we had what Merck required.”
Of course, IBM also has cross-segment infrastructure solutions (software, hardware, middleware, services) upon which to run segment specific applications.
Also in the biopharma industry, IBM has broadly identified opportunities in biomarker (imaging and omic) management, in silico R&D, data integration, and regulatory compliance. Imaging is a particularly hot area.
The IBM Healthcare and Life Sciences organization organized four Biobank Summits and four Imaging Biomarker Summits, led by Hehenberger, as part of its more collaborative approach to develop solutions to industry problems and products IBM can sell. These meetings drew attendees from FDA, large pharma, and technology vendors including GE, Siemens, and Philips.
Hehenberger says drug approval is often delayed—even prevented—because of insufficient imaging information. Sometimes he says, images have been collected but cannot be located.
Avastin and ImClone’s Erbitux are famous cases, he says. “[ImClone] couldn’t convince the FDA because they didn’t find the data. It was a good drug. I mean they may have made some mistakes also in their design of the clinical trials, but imaging was an important factor. It’s important that the image quality [and] then the management of the images [be handled effectively] in a regulated environment.”
Big Blue expects the growth in imaging for diagnostics and drug development to drive a huge market for technology enablers. Initially, it has steered clear of PACS imitation, focusing instead on clinical trials, and is even preparing a hosted service, which will manage images for clinical trial sponsors. As is increasingly typical, parts of the solution came from IBM’s IT infrastructure portfolio, part from IBM Research efforts, and extensive customer involvement.
“I’m not saying we have solved all the problems,” says Hehenberger. “There’s a challenge of making things easy to use, of not deviating too much from the environment they’re used to, but still sending those images to the CRO and then to the pharmaceutical company. This makes it possible for the imaging department and other investigators to study the images and monitor what’s happening. There’s a lot of communication going on and those are big files, so, it’s very stressful for the IT infrastructure.”
“We have [also] recognized the importance of modeling and simulation and we’re going to do what we did for imaging biomarkers.” His group is planning conferences in Japan and New York by year’s end.
Things have changed since 1989, when Hehenberger’s priority in selling to pharma was to make Macs and PCs behave as dumb terminals to preserve the role of mainframes and their applications. “I’m very happy for this second opportunity,” he says, his exuberance tempered by past realities and seems to be shared by today IBMers. They don’t forget the struggle.
Another interesting offering is IBM’s Unstructured Information Management and Mining Solution (UIM). “We have now four implementations in the industry of what we call business intelligence workbench. The idea is to integrate sources of unstructured data and to annotate them and, again, have an open architecture. So, where we don’t have the best algorithm ourselves, we can actually import outside ideas and algorithms for annotation. You can look at chemicals, biomarkers, genes, proteins, diseases, symptoms, etc.,” says Hehenberger.
IBM Healthcare and Life Sciences’ portfolio is extensive, powerful, and to a considerable degree, customer and partner driven. For example, IBM and Oracle jointly offer the Life Sciences Hub, a platform and suite of applications to manage clinical trial and other biomedical data.
Svinte notes IBM solutions should help drive business innovation. So for example, SCORE shouldn’t just keep a company in compliance but also provide intelligence needed to inform a company’s strategy.
Services are a big piece of Healthcare and Life Science’s business. Recently, for example, IBM extended its contract with AstraZeneca to manage its IT infrastructure. The new deal is for $1.4B over seven years; parts of the new collaboration extend to research, not just IT.
“IBM is partnering with AstraZeneca, Karolinska Institute (KI), and other academic institutions in the Stockholm Brain Institute (SBI), an example of a public/private partnership. The emergence of such PPP’s is a strong global trend. SBI’s research is focused on the basic understanding of cognition and learning. Cognitive dysfunction is the cause of many important diseases such as Alzheimer’s, Parkinson’s, ADHD, Schizophrenia, etc.,” says Hehenberger.
“In addition to the analysis of molecular and clinical data, functional MRI and PET (via Siemens’ HRRT instrument with the industry’s highest possible resolution) imaging studies, SBI is using computer simulations based on IBM’s Blue Gene platform. The objective is to get an understanding of neuroscience disease mechanisms and to find new biomarkers. The SBI wants to get research results that can be used to develop new medical treatments and can be translated into benefit, ultimately, for the patient.” he says.
Svinte says the range and scope of IBM services keeps expanding. For example, IBM has for some time offered supply chain management and procurement services. In any case, IBM’s culture and mandates are such now that his organization can bring all internal parties—services, products, IBM Research—to bear on client engagements.
Over $30B has been spent to fund IBM R&D in the past five years. No doubt big chunks go to support the nuts and bolts of infrastructure hardware and software development.
But this is also the land of Blue Gene, where computational brawn and elegance are combined to tackle giant and tiny problems. It’s also Ajay Royyuru’s workplace (and toy box).
Royyuru is the senior manager of IBM Research’s Computational Biology Center (CBC) at Watson Research Center. (See columnist Michael Greeley’s article, “What’s New at Big Blue,” Bio•IT World, March 2005) He leads 35 researchers digging into projects covering bioinformatics, structural biology, functional genomics, systems biology, and medical informatics. Royyuru is himself the lead scientist for IBM on the Genographic Project.
His training includes a master’s degree in biophysics, a Ph.D. in molecular biology in India, and postdoctoral study in structural biology at Memorial Sloan-Kettering Cancer Center, providing a breakthrough animal model for HIV research. Prior to joining IBM Research in 1998, Royyuru spent two years developing structural biology software at Accelrys.
IBM’s CBC covers a truly surprising array of research projects, from mining and modeling of biological networks (Gustavo Stolovitzky) to structural studies of rhodopsin and its methods of activation (Mike Pitman); and from high-performance computing for medical imaging and neuroscience (Charles Peck) to genetic regulatory motifs (Isidore Rigoutsos).
Although much of this sounds very academic, all of the work is embedded in IBM’s solutions-for-real-problems mindset. Many projects are brought to IBM by clients who pay to put Big Blue to work. As outlined by Joseph Jasinski, IBM distinguished engineer and program director, Healthcare & Life Sciences, the research effort has five strategic focuses:
- Improve effectiveness, safety, and cost of clinical care and public health through better diagnosis, treatment, and operational efficiency
- Improve operational efficiency and customer relationship capabilities of health care payers and consumers
- Develop new, more effective, drugs faster and cheaper
- Translate molecular biology research into medical care
- Understand biological systems with predictive models
“The IBM Research business model has changed dramatically,” says Jasinski. “80% of the work [Ajay’s] computational biology center does is with clients and collaborators. Why does he do that? Not to work on today’s issues, but [to identify] the massive emerging challenges on a 3-to–5 year horizon… That is such a huge advantage to what we do and what we are able to offer to our clients.”
Some terrific science is being done as well. Isidore Rigoutsos, manager, Bioinformatics and Pattern Discovery, is leading a computational effort to examine non-coding (junk) DNA. His work suggests there are a huge number of conserved intronic sequences called pyknons that have regulatory roles. Moreover, while they cluster in similar functional areas across organisms, they tend to have different actual sequences.
Such insights could have profound impact on understanding animal models of disease and perhaps explain why they are often poor predictors of human response. His work has identified pyknons implicated in colon cancer. And the clever pattern recognition algorithms and techniques he has developed are likely eventually to be embedded in tools.
Gustavo Stolovitzky, manager, functional genomics & systems biology, has multiple research interests. Some efforts—his work with the DREAM Project (Dialogue on Reverse Engineering Assessment and Methods) and heart modeling and simulation—are fairly visible. Other projects are still largely confidential but tantalizing, such as some exciting preliminary work on nanopore DNA sequencing.
There are probably many proprietary programs kept under the close eye of sponsoring clients. This is part of the new IBM Research mission. Curiosity alone is generally not enough.
Apart from the human brain power, the other jewel in IBM Research is Blue Gene, or rather the family of Blue Gene supercomputers, announced in 1999, and whose predecessor Deep Blue defeated chess legend Garry Kasparov in 1997. Blue Gene/L debuted in 2004 and grew in performance to the hundreds of teraflops. This is the version that the IBM CBC used.
Yet just down the hall in the Watson Lab is a Blue Gene/P, which when fully provisioned with 256 racks, will achieve 3.56 petaflops. This latest computational beast has had virtually everything optimized—cabling, cooling, power, etc,—and will permit biological simulations on a much faster and deeper scale. Of course, the line won’t stop there. Plans call for a 100-petaflop Blue Gene/Q to debut around 2010.
Clearly not everyone can afford (or has the expertise to run) such a behemoth. IBM says the effort required to program Blue Gene has been vastly simplified, and it even sells access to Blue Gene as an “on demand” service to clients such as Gene Network Sciences.
Clearly IBM has come a long way. The chastened giant stumbled but found the grit to change and regain its stride in what is arguably a tougher, coarser, less forgiving business environment. Says Svinte: “IBM’s strategy, not just Healthcare and Life Sciences, has three planks—open technologies and high value solutions; integration innovation; and becoming a premier global integrated enterprise. That’s driving all of our initiatives.”
IBM CEO Survey Shows Hunger for Change in Life Sciences
“Change” is not merely a common refrain on the presidential election campaign trail. It’s also what dozens of life science CEOs are hankering for, according to preliminary findings from the latest IBM CEO survey.
Michael Svinte shared some of those findings with Bio•IT World at a briefing during the Drug Industry Association conference (DIA) this June in Boston.
Every two years, IBM conducts a comprehensive CEO study, surveying hundreds of industry chief executives. For the 2008 survey, IBM spoke with 1,130 CEOs, including 40 in the life sciences and pharma arena. (Eight of the CEOs came from the Americas, 46 percent from Europe/Middle East/Africa and 34 percent from Asia-Pacific.)
Svinte says the results, which were published in a full white paper, are important in allowing IBM to allocate its resources into appropriate areas of need.
There were five principal takeaways from the life science CEO responses, according to Svinte. Chief among them was a hunger for change. 79 percent of the life science CEOs anticipated major change in the next few years. Talent concerns, regulatory issues, and market factors were the most commonly cited factors.
But 25 percent of the CEOs polled said they were struggling to manage change within their organizations. One American big pharma CEO said, “We know we need to change faster, but are we adaptive enough? There is a lot of skepticism internally with regard to our change capabilities.”
The other key lessons Svinte and colleagues extracted from the survey include:
• Innovation beyond customer imagination: CEOs were “bullish” about the increase in consumer purchasing power. “Patients will be more involved because they’re better informed,” said one European pharma CEO. The survey showed investment in this area projected to grow by 43 percent in the coming years, but less for more sophisticated consumers. Svinte says life science companies face a huge challenge “engaging the broader ecosystem,” (similar to IBM’s evolution in recent years), as they pursue new markets, new business operations, and new operations.
• Global integration: 90 percent of life science CEOs questioned are entering new markets, but they are wary of legislative regulation and intellectual property concerns. As with the larger study, insufficient talent is a serious obstacle, they said.
• Business model innovation: Three quarters of the life science CEOs are pursuing “extensive business model innovation over the next three years.” Here, the emphasis among the CEOs is on enterprise model innovation. “Our future is in convergent technologies, where we combine different aspects of our business with technology, for example, combining a technological component with medication,” said one European pharma CEO. Surprisingly perhaps, fewer than 20 percent of the CEOs are entertaining industry model innovation. IBM expects to see merging capabilities among pharma, health care and diagnostic companies as new industry models emerge.
• Social responsibility: 62 percent of life science CEOs reacted favorably to exhibiting social responsibility. This primarily impacts the manufacturing space and medical by-products, yet one quarter of CEOs doubted that improved social responsibility would enhance their business. -- Kevin Davies
This article appeared in Bio-IT World Magazine.
Subscriptions are free for qualifying individuals. Apply Today.