Abstract

How do you search for something when you don’t know what size it is, what it looks like, or where it is? One way of gaining new understanding about the lifecycles of RNA viruses (a large proportion of all known viruses) is to search for previously undetected regions in the viral RNA where there is unexpectedly high nucleotide conservation – suggesting that mutations in those regions are for some reason lethal to the virus. Because some such regions lack analogues in the host (i.e. are specific to the virus) and because there are high barriers to mutation, such regions are strong candidates to be antiviral drug targets.

Many ways of looking for high conservation presume some foreknowledge of the motifs we are looking for – but to find truly novel biology such methods alone are inadequate. I shall describe a way of approaching this problem, which turns out to be related to the properties of random walks. I shall show how we have applied this method to analysing HIV and predicted previously undescribed structure in the RNA with important consequences for viral function. In doing so I shall take you from open problems in probability theory, through information theory and molecular biology, to applications in clinical medicine.