Bioinformatics is currently a very active research area, mainly because
of the recent advancements in genomics and related technologies.
"Smart" methods for automating or assisting analysis and understanding
are required by the size and complexity of these data. Machine learning
techniques (such as unsupervised analysis via clustering and self-organizing
maps, subspace methods, classification with support vector machines,
with neural networks, with decision trees, and so on) provide valuable
tools for these tasks.
In particular, we have focused on the analysis of DNA microarray data
also through a partnership with the Advanced Biotechnology Center of the
University of Genova, which provides a facility for the production of
DNA microarrays. We have studied the application of several types of
classifiers to disease discrimination, class discovery, differential
diagnosis. One research contribution is in the field of ensemble machines,
a technique to combine several classifiers into a composite one. This
allows a pool of (possibly heterogeneous) two-class classifiers to
tackle multi-class classification problems.
We are also studying a technique to automate and optimize the design of
oligonucleotide microarrays to be applied to the typing of the human HLA
system, a central immunologic problem in transplantology currently tackled
with other, more expensive approaches.