Toward Pangenome Analysis: the graph-based approach скачать в хорошем качестве

Toward Pangenome Analysis: the graph-based approach

Title: Toward Pangenome Analysis: the graph-based approach Speaker: Mirko Coggi Area/Topic/Keywords: Pangenomics, Genome Graphs, Hardware Acceleration Short Abstract: Since its discovery, the genome become the primary key to unlocking the specific nature of each organism. Next Generation Sequencing pushes this quest for specificity even further, generating a vast amount of data that reveals how genome variability, both inter- and intra-individual, has to be the first-class aspect in the future evolution of the field. However, standard sequence-based analysis pipelines fail to keep up with genetic variability because of data redundancy and the inability to resolve structural variations. Graphs are the prominent data structures to tackle these shortcomings, thanks to their ability to natively encode variations as alternative paths. Finding specific sequences in graph-based reference data is achieved through the sequence-to-graph alignment process, implemented by a growing family of compute-intensive algorithms that find the path in the graph that best represents the query sequence. Each of these algorithms bets on a fragile balance between accuracy – implementing cyclic graphs support, traceback computing, and standardization of input/output file formats – and performance – relying on multi-threading approaches, heuristic strategies, and heterogeneous computing. This talk reviews some open-source sequence-to-graph alignment tools in the State of the Art, benchmarking their features and how respective design choices impact the accuracy and performance of the alignment process, concluding with clear guidelines needed to drive the evolution of this fertile line of research.