Predição da estrutura 3D de proteínas mimetizando o ambiente ribossômico

Abstract

Protein structure prediction from just the amino acid sequence continues to be a major challenge in structural bioinformatics. If at all possible, prediction needs to be accurate and fast. In this project, it is proposed and tested the effects of cotranslation within an ideal ribosomal channel model in protein structure prediction using classical molecular dynamics and replica-exchange molecular dynamics simulations. An ideal ribosomal channel model was built, different translation speeds were used and compared the results to control simulations. Different translation speeds were tested to verify their influence on predictions, and the best results were observed at translation speeds between 80 and 200 ps. The quality of the predicted models were as low as 0.3 Å and 1.0 for the RMSDs and GDT-TS parameters, respectively, for simulations of just 50 ns. Overall, the use of this approach to protein structure prediction has successfully produced native and near-native structures in three of the four proteins investigated, thus reaching accuracy and speed as expected. As a conclusion, using cotranslation within an IRCM is a promising approach to predict native-like 3D structures of mini-proteins successfully. Improvements to the methodology should allow the prediction of 3D structures of larger proteins of biological and biomedical interest.