IRB Barcelona and Anaxomics, among best in the world in bioinformatics to diagnose disease

By Biocat

The biosciences are generating a huge amount of data at an unprecedented speed, making the extraction of useful, reliable data ever more difficult and complex. In May, IBM Research and Philip Morris International (PMI) launched the Improver project (Industrial Methodology for Process Verification in Research) in Boston (United States) in order to challenge the best biocomputational researchers from around the world to demonstrate the functionality of their methods to exploit genomic information for predictive and clinical ends, and that their results are reliable and verifiable.

The first challenge posed under the Improver project, called Diagnosis Signature, consisted in separating and classifying groups of patients for four diseases (psoriasis, multiple sclerosis, chronic obstructive pulmonary disease and lung cancer) and exploiting genomic data from clinical samples. The team led by IRB Barcelona and Anaxomics Biotech finished fourth in the world, in a competition among 54 teams mainly from Europe and the United States. The presentation of the project was published in Nature Biotechnology.

According to Dr. David Rossell, head of the IRB Barcelona Biostatistics and Bioinformatics Platform, “these international challenges are highly effective as proof of concept and as a way to expose the world to the most effective, cutting-edge techniques to do this."

The method: incorporating biologic information into statistical algorithms

Anaxomics Biotech contributed information published on the proteins involved in each of the diseases. Dr. Patrick Aloy, head of the joint IRB Barcelona-BSC computational biology program, added to this biological documentation with non-published data on other molecules involved. Afterwards, Rossell integrated all the biological information into his probability prediction methods.

"Competition fosters the creation of a strong body of scientific evidence for the biomedical community, both researchers and doctors, and that adds value and lends credibility to biocomputational techniques and their possible industrial applications,” commented Aloy.

The first Improver challenge highlighted the predictive capabilities of biocomputation, as most participants were able to diagnose patients with a certainty level of nearly 90%.

The organizers expect to continue the Improver project over four years. For now, they have announced that the second challenge will be launched in 2013.