2021 Mollie Holman winner: Dinithi Sumanaweera

27 June 2022


The Mollie Holman Medal was established in 1998 and is named after the late pioneering physiologist Emeritus Professor Mollie Homan AO, in honour of her significant contributions to science and education. Through this award, the University seeks to acknowledge and reward its doctoral candidates for both the quality of their thesis and the quality of their research.  The awards are among the highest academic honours the University bestows, and mark the recipient as the researchers of the highest order.

Why are protein sequence alignments so important?

Proteins control all aspects of cellular and molecular processes in organisms. They are composed of amino acids, and experimental methods that characterise proteins as sequences of amino acids form an important data stream for molecular biology research. Extant proteins diverge from their common ancestors while tolerating considerable variation in their amino acid sequences. Therefore, inferring trustworthy evolutionary relationships between their sequences is a challenging computational task and, when performed reliably, provides a powerful way to reason about the macromolecular consequences of evolution, thus forming a crucial first-step to drive many downstream research studies in biology and medicine.

Dinithi’s innovations

Dinithi’s thesis makes innovative contributions to the theory and practice of protein sequence alignment. It develops a Bayesian framework that combines inductive inference with information theory that not only provides the ability to search for a single best alignment, but also supports the exploration of competing alignments and visualisation of the entire landscape of all possible alignment relationships. No existing alignment method achieves this. The thesis also developed new stochastic Markov matrices of amino acid substitution that outperform all widely used substitution matrices. Importantly, this thesis overcomes a key shortcoming in the field by unifying statistical models for amino acid substitution with those that account for insertions and deletions. This is the first such demonstration that such complete models are feasible.