Minimum Factorization Agreement of Spliced ESTs

Producing spliced EST sequences is a fundamental task in the computational problem of reconstructing splice and transcript variants, a crucial step in the alternative splicing investigation. Now, given an EST sequence, there can be several spliced EST sequences associated to it, since the original EST sequences may have different alignments against wide genomic regions. In this paper we address a crucial issue arising from the above step: given a collection C of different spliced EST sequences that are associated to an initial set S of EST sequences, how can we extract a subset C′ of C such that each EST sequence in S has a putative spliced EST in C′ and C′ agree on a common alignment region to the genome or gene structure? We introduce a new computational problem that models the above issue, and at the same time is also relevant in some more general settings, called Minimum Factorization Agreement (MFA). We investigate some algorithmic solutions of the MFA problem and their applicability to real data sets. We show that algorithms solving the MFA problem are able to find efficiently the correct spliced EST associated to an EST even when the splicing of sequences is obtained by a rough alignment process. Then we show that the MFA method could be used in producing or analyzing spliced EST libraries under various biological criteria.