P2-0209 — Annual report 2011
1.
Learning qualitative models from numerical data

The paper describes Pade, a new method for qualitative learning which estimates partial derivatives of the target function from training data and uses them to induce qualitative models of the target function. We formulated three methods for computation of derivatives, all based on using linear regression on local neighbourhoods. The methods were empirically tested on artificial and real-world data. We also provide a case study which shows how the developed methods can be used in practice.

COBISS.SI-ID: 8324436
2.
SNPsyn : detection and exploration of SNP-SNP interactions

We developed an interactive tool for the analysis of GWAS data (genome-wide association studies). Data on over one million SNPs (single nucleotide polymophism) for few thousands cases and healty controls are gathered in a typical GWAS study. The SNPsyn application allows a user-friendly analysis of the association of individual SNPs and identification of synergistic pairs of SNPs associated with disease. We derived and use an upper bound on synergy to speed-up the combinatorial search and scoring of pairs of SNPs.

COBISS.SI-ID: 8352596
3.
Analysis of alternative splicing associated with aging and neurodegeneration in the human brain

Comparing samples from healthy individuals ranging from 16 to 102 years old with samples from diseased individuals, we uncovered a striking similarity in the changes in gene expression patterns associated with aging and the neurodegenerative diseases. We evaluated changes in alternative splicing in great detail, which showed that changes associated with aging largely affected genes associated with cellular metabolism, while disease-specific changes were associated with genes involved in neuron-specific function. Finally, we found changes in the expression of several genes coding for RNA binding proteins, which are likely responsible for at least part of the observed alterations in splicing.

COBISS.SI-ID: 8549204