Protein evolution is most commonly studied by analyzing related protein sequences and generating ancestral sequences through Bayesian and Maximum Likelihood methods, and/or by resurrecting ancestral proteins in the lab and performing ligand binding studies to determine function. Structural and dynamic evolution have largely been left out of molecular evolution studies. Here we incorporate both structure and dynamics to elucidate the molecular principles behind the divergence in the evolutionary path of the steroid receptor proteins. We determine the likely structure of three evolutionarily diverged ancestral steroid receptor proteins using the Zipping and Assembly Method with FRODA (ZAMF). Our predictions are within ∼2.7 Å all-atom RMSD of the respective crystal structures of the ancestral steroid receptors. Beyond static structure prediction, a particular feature of ZAMF is that it generates protein dynamics information. We investigate the differences in conformational dynamics of diverged proteins by obtaining the most collective motion through essential dynamics. Strikingly, our analysis shows that evolutionarily diverged proteins of the same family do not share the same dynamic subspace, while those sharing the same function are simultaneously clustered together and distant from those, that have functionally diverged. Dynamic analysis also enables those mutations that most affect dynamics to be identified. It correctly predicts all mutations (functional and permissive) necessary to evolve new function and ∼60% of permissive mutations necessary to recover ancestral function.

Background:
Many pharmaceutical drugs are known to be ineffective or have negative side effects in a substantial proportion of patients. Genomic advances are revealing that some non-synonymous single nucleotide variants (nsSNVs) may cause differences in drug efficacy and side effects. Therefore, it is desirable to evaluate nsSNVs of interest in their ability to modulate the drug response.

Results:
We found that the available data on the link between drug response and nsSNV is rather modest. There were only 31 distinct drug response-altering (DR-altering) and 43 distinct drug response-neutral (DR-neutral) nsSNVs in the whole Pharmacogenomics Knowledge Base (PharmGKB). However, even with this modest dataset, it was clear that existing bioinformatics tools have difficulties in correctly predicting the known DR-altering and DR-neutral nsSNVs. They exhibited an overall accuracy of less than 50%, which was not better than random diagnosis. We found that the underlying problem is the markedly different evolutionary properties between positions harboring nsSNVs linked to drug responses and those observed for inherited diseases. To solve this problem, we developed a new diagnosis method, Drug-EvoD, which was trained on the evolutionary properties of nsSNVs associated with drug responses in a sparse learning framework. Drug-EvoD achieves a TPR of 84% and a TNR of 53%, with a balanced accuracy of 69%, which improves upon other methods significantly.

Conclusions:
The new tool will enable researchers to computationally identify nsSNVs that may affect drug responses. However, much larger training and testing datasets are needed to develop more reliable and accurate tools.