Resumen de Large-scale comparative bioinformatics analyses
One of the main and most recent challenges of modern biology is to keep-up with growing amount of biological data coming from next generation sequencing technologies. Keeping up with the growing volumes of experiments will be the only way to make sense of the data and extract actionable biological insights. Large-scale comparative bioinformatics analyses are an integral part of this procedure. When doing comparative bioinformatics, multiple sequence alignments (MSAs) are by far the most widely used models as they provide a unique insight into the accurate measure of sequence similarities and are therefore instrumental to revealing genetic and/or functional relationships among evolutionarily related species. Unfortunately, the well-established limitation of MSA methods when dealing with very large datasets potentially compromises all downstream analysis. In this thesis I expose the current relevance of multiple sequence aligners, I show how their current scaling up is leading to serious numerical stability issues and how they impact phylogenetic tree reconstruction. For this purpose, I have developed two new methods, MEGA-Coffee, a large scale aligner and Shootstrap a novel bootstrapping measure incorporating MSA instability with branch support estimates when computing trees. The large amount of computation required by these two projects was carried using Nextflow, a new computational framework that I have developed to improve computational efficiency and reproducibility of large-scale analyses like the one carried out in the context of these studies.
