Big-data approaches to protein structure prediction

See allHide authors and affiliations

Science  20 Jan 2017:
Vol. 355, Issue 6322, pp. 248-249
DOI: 10.1126/science.aal4512

You are currently viewing the summary.

View Full Text

Log in to view the full text

Log in through your institution

Log in through your institution


A protein's structure determines its function. Experimental protein structure determination is cumbersome and costly, which has driven the search for methods that can predict protein structure from sequence information (1). About half of the known proteins are amenable to comparative modeling; that is, an evolutionarily related protein of known structure can be used as a template for modeling the unknown structure. For the remaining proteins, no satisfactory solution had been found. On page 294 of this issue, Ovchinnikov et al. (2) used recently developed methodology for predicting intraprotein amino acid contacts in combination with protein sequences from metagenomics of microbial DNA to compute reliable models for 622 protein families, and discovered more than 100 new folds along the way. The fast-paced growth of metagenomics data should enable reliable structure prediction of many more protein families.