Below is a guest blog from Ketan Paranjape, director of personalized medicine at Intel. He is speaking tomorrow, May 22, at the Big Data in BioMedicine Conference at Stanford University starting at 2 p.m. You can watch the live stream of the conference here and get an inside look at what’s going on with big data and bio IT. Follow the #bigdatamed hashtag on Twitter during the three-day conference.
Over the weekend I was perusing the latest edition of the New York Magazine and read this article about a cancer doctor who lost his wife to cancer. After I put my tablet down, I hugged my wife, grabbed the “honey do list” from her hand and got to work. Whether I was comparing the right mix of fertilizers for my lawn at the neighborhood home improvement store or driving my son to his tennis practice, all I could think about was, “what could we have done to prevent what happened and also help the 9 million or so cancer patients worldwide?”
Here at Intel, we are trying our best to answer that very question. We have laid out a vision for delivering “Personalized Medicine at the touch of a button…Everywhere…Everyday…and for Everyone…” I suppose you need an app for everything these days.
But you would wonder how could a hardware company that makes “chips” for computers make a difference? The answer from our vantage point is Big Data Analytics. If we can create an ecosystem of hardware, software and services players in healthcare and life sciences that generate, manage, share, and analyze data across the multiple silos of payers, government, providers, clinics, pharma, clinical trials, genomics, and wellness data, we should be able to help the key decision makers with the right amount of well curated information.
So starting out with what we do best, we optimized software programs for analyzing data. We were able to cut the run time of Broad’s Genome Analysis ToolKit (GATK) from three days to one day and speed up Ayasdi’s Cure software by 400 percent. These reductions in run times and performance improvements help speed up the number of tests you can do and help the researcher or clinician get to an answer quicker.
We then built appliances – highly optimized hardware and software solutions that target specific genomics or clinical problems. The appliance with Dell reduces the RNA seq run times from seven days to three hours and the Genalice appliance can map 42 whole genomes in 18 hours. As you start running out of machines or clusters in our backyard, we go to the cloud. Working with Novartis, Amazon Web Services (AWS), Cycle Computing and MolSoft, we were able to provision a fully secured cluster of 30,000 CPUs and completed screening 3.2 million compounds in about nine hours compared to 4-14 days on existing resources.
We have also embraced the power of both Hadoop and in-memory analytics. Working with Nextbio (now part of Illumina), we were able to get a 90 percent gain in throughput and 6x data compression enabling researchers to discover biomarkers and drug targets by correlating genomic data sets. Working with SAP and Charite we built a ‘real-time’ cancer analysis platform on SAP-HANA that analyzed patient data in seconds compared to two days. BTW, this data set had 3.2M data points per patient and up to 20 TB of data/patient.
These are just a subset of projects we have completed working with our broad range of ecosystem partners. For more details please visit – www.intel.com/healthcare/bigdata or drop me a line – Ketan.firstname.lastname@example.org. I’m also speaking tomorrow, May 22, at the Big Data in BioMedicine Conference at Stanford University starting at 2 p.m. You can watch the live stream of the conference here
By the way, regarding the fertilizer we decided on a good Nitrogen mix (2 lbs of Nitrogen per 1000 sq. feet) and my son moved from level 1 to level 1.5 in his class.
What questions do you have?