Inferring repeat-protein folding features from sequences

GALPERN, EZEQUIEL A.; FERREIRO, DIEGO U.

Buenos Aires

Simposio; 6to Simposio Argentino de Jovenes Investigadores en Bioinformatica; 2021

The coding space of protein sequences is shaped by evolutionary constraints set byrequirements of function and stability. Repeat proteins are made with tandem copiesof similar amino acid stretches that fold into elongated architectures. Due to itssymmetries, they are a unique system to model how evolutionary constraints at thesequence level can impact tertiary structure, folding and function. In this work wedepart from pairwise identity patterns emerged from a curated Ankyrin 150000sequence dataset. We combine an inverse Potts model scheme with an explicitmechanistic model of duplications and deletions of entire repeats for calculating theevolutionary parameters of the system, such that the observed identity patterns canbe reproduced. We use the evolutionary energy obtained to input a folding toy modelfor repeat proteins, a coarse-grained 1D ising model where each spin corresponds to aprotein sequence fragment and has to be folded or unfolded. We perform Monte Carlosimulations of the model using only sequence information and we get thermalunfolding curves that are compatible with experimental ones of several proteins. Weperform a large scale analysis of the dataset and we develop a predictor of severalfolding mechanism features for natural or designed Ankyrin proteins.