I unfortunately don’t have the detailed picture, but only several layman’s view of the problems they were working on.
From what I understood, a lot of their research involved pattern recognition in strings of the size of about 4-6 GiB, combinatorics of smaller patterns, classification across multiple samples etc. I kept thinking these are elegantly solved problems in other domains, and if one were to possess both the skills of algorithms and the knowledge of what is it they are looking for, then it can click for a win.