Predicting Variants of Alternative Splicing from NGS data without the Genome

Known approaches to determine the gene structure, due to alternative splicing (AS), rely on some forms of spliced alignment of the transcripts against the genomic sequence of the gene producing the transcripts. Anyway, the abundance of data arising from next-generation sequencing technologies allows a new approach, proposed in this paper, that does not require any kind of alignment of the transcripts against the genome. We formalize the underlying computational problem, proving the soundness of our approach from both a theoretical and experimental point of view.