The field of Bioinformatics faces several challenges at present: (i) a data explosion driven by novel sequencing technologies, (ii) the need for better, more accurate, and thus, more compute-intensive models, and (iii) the need to improve bioinformatics software quality as well as scalability. In this talk, I will outline, by example of evolutionary biology, how we can overcome some of these challenges, for instance by designing appropriate parallel I/O procedures and data distribution algorithms. I will also discuss the pros and cons of accelerator deployment as well as adaptation to specific supercomputer architectures for these applications. Subsequently, I will outline typical compute patterns and hardware requirements of large-scale biological data analyses that do not always require classic supercomputers. Thereafter, I will critically discuss some of the substantial problems related to code quality in Bioinformatics that we have recently identified. I will conclude with a quote by Donald Knuth: "Biology easily has 500 years of exciting problems to work on." |
;){kind=link}