Bioinformatical Analysis of Alternative Splicing

Abstract

Splicing is a natural process happening in every living cell. As a fundamental process, all genes undergo splicing before tuning into a functional molecule such as proteins. However, the splicing process is still under active research in order to completely understand how it is regulated. As of now, the splicing machinery is characterized in what we call the spliceosome, but a complete understanding of it is still under active research. The discovery of alternative splicing set a breakthrough in molecular biology. This process allows a simple gene to turn into more than one protein by creating different transcripts or isoforms. The transcriptome is the whole set of known isoforms that have been characterized through research and experimental procedures. Still, many isoforms might remain unknown. Through research, alternative splicing has been shown to be responsible for the development of different pathologies, including cancer. On the other hand, alternative splicing provides a novel mean to characterize potential biomarkers for drug resistance or survival. Due to this, alternative splicing is constantly being studied in order to completely understand its functioning mechanism.

With the development of sequencing techniques, it has been possible to develop methods to quantify the different isoforms present in specific samples or conditions. From the method, developed by Frederick Sanger, also known as Sanger sequencing, to the development of new protocols, like third generation sequencers, vast amounts of data have been generated. Even though, sequencing has taken a lead role in the study of alternative splicing, other platforms like junction arrays have been developed to study such phenomenon. Affymetrix recently developed the Clariom microarray, which seems to be the most up to date array to identify splicing events.

Most of the available algorithms to identify and quantify alternative splicing events, provide more than one figure of merit per event and does not take into account a coherence in such events. For example, in a cassette exon, not only the skipped exon should show a change in expression, but the flanking junctions should display a similar behavior but in opposite directions. In order to take this into account, a novel method to identify, classify and state statistical significance of splicing events has been developed. EventPointer allows users to identify alternative splicing events and provide the statistical significance of such events. The algorithm can be applied to data from both microarrays or RNA-Seq. Also, EventPointer generates files that can be loaded into genome browsers to ease the interpretation of the results and the desing of primers for standard PCR validations.

The performance of EventPointer has been tested in two independent experiments using both platforms. The overall results show a promising validation rate in both technologies. EventPointer, also estimates the percent spliced index for every detected event and not only skipping exons, as most of the available software. The results, obtained through end-point PCR demonstrate that the estimated $\Psi$ values, provided by EventPonter, are highly correlated with the experimental results. EventPointer shows an improved method to identify and quantify alternative splicing events.

A comparison between microarrays and RNA-Seq, in their ability to identify alternative splicing events was performed using the same experimental data from three different cell lines treated with a drug that severely affects the splicing machinery. The results show that RNA-Seq is the most flexible and trustable platform for the identification of splicing events, but microarrays are a viable option to analyze alternative splicing due to reasons of cost and convenience. Microarrays can be an alternative when compared to shallow sequencing.