アブストラクト
| Date | 10:30am March 20, 2023 |
| Speaker | Pascal Hingamp Aix-Marseille University, France |
| Title | Environmental genomics: bioinformatics methods to estimate biodiversity from metagenomes |
| Abstract |
Methods to extract phylogenetic data from metagenomics datasets are evolving rapidly. These changes are due to two opposing consequences of DNA sequencer’s throughput acceleration. First, metagenomic datasets are routinely reaching volumes so vast, that the classical sequence alignment algorithms that have dominated bioinformatics since the 70s are now proving to slow to cope, even when factoring in concomitant increase in compute capacity. Second, the very significant increase in available reference genomes means the exquisite complexity of gapped mismatched alignments can now be replaced by much simpler and lightening fast searches for short exact matches. Such novel alignment-free k-mer algorithms happen to nicely accommodate the shortness of reads characteristic of second generation DNA sequencing, going as far as allowing direct assembly-free taxonomic estimates of raw reads. I will present currently used methods for taxonomic evaluation of metagenomes, both supervised (dependent on external reference databases) and unsupervised (eg binning/compositional methods), using either filtered marker gene sets extracted from the metagenomes or unfiltered full scale environmental shotgun datasets.
|