Sequence space (evolution)

Abstract: Many mutations disappear from the population because they impair protein function and/or stability. Thus, amino acid positions that are essential for proper function evolve more slowly than others, or in other words, the slow evolutionary rate of a position reflects its importance. Con- Surf (http://consurf.tau.ac.il), reviewed in this manuscript, exploits this to reveal key amino acid positions that are im- portant for maintaining the native conformation(s) of the protein and its function, be it binding, catalysis, transport, etc. Given the sequence or 3D structure of the query protein as input, a search for similar sequences is conducted and the sequences are aligned. The multiple sequence alignment is subsequently used to calculate the evolutionary rates of each amino acid site, using Bayesian or maximum-likelihood algorithms. Both algorithms take into account the evolution- ary relationships between the sequences, reflected in phylo- genetic trees, to alleviate problems due to uneven (biased) sampling in sequence space. This is particularly important when the number of sequences is low. The ConSurf-DB, a new release of which is presented here, provides precalcu- lated ConSurf conservation analysis of nearly all available structures in the Protein DataBank (PDB). The usefulness of ConSurf for the study of individual proteins and mutations, as well as a range of large-scale, genome-wide applications, is reviewed.

Abstract: To reduce experimental effort associated with directed protein evolution and to explore the sequence space encoded by mutating multiple positions simultaneously, we incorporate machine learning into the directed evolution workflow. Combinatorial sequence space can be quite expensive to sample experimentally, but machine-learning models trained on tested variants provide a fast method for testing sequence space computationally. We validated this approach on a large published empirical fitness landscape for human GB1 binding protein, demonstrating that machine learning-guided directed evolution finds variants with higher fitness than those found by other directed evolution approaches. We then provide an example application in evolving an enzyme to produce each of the two possible product enantiomers (i.e., stereodivergence) of a new-to-nature carbene Si–H insertion reaction. The approach predicted libraries enriched in functional enzymes and fixed seven mutations in two rounds of evolution to identify variants for selective catalysis with 93% and 79% ee (enantiomeric excess). By greatly increasing throughput with in silico modeling, machine learning enhances the quality and diversity of sequence solutions for a protein engineering problem.

Abstract: The biological activity of a protein typically depends on the presence of a small number of functional residues. Identifying these residues from the amino acid sequences alone would be useful. Classically, strictly conserved residues are predicted to be functional but often conservation patterns are more complicated. Here, we present a novel method that exploits such patterns for the prediction of functional residues. The method uses a simple but powerful representation of entire proteins, as well as sequence residues as vectors in a generalised 'sequence space'. Projection of these vectors onto a lower-dimensional space reveals groups of residues specific for particular subfamilies that are predicted to be directly involved in protein function. Based on the method we present testable predictions for sets of functional residues in SH2 domains and in the conserved box of cyclins.