2009 | Faruck Morcos, Andrea Pagnani, Bryan Lunt, Arianna Bertolino, Debora S. Marks, Chris Sander, Riccardo Zecchina, José N. Onuchic, Terence Hwa, Martin Weigt
A computationally efficient implementation of Direct Coupling Analysis (DCA) has been developed to predict residue contacts in protein domains based on sequence information. This method, called mfDCA, is significantly faster than the previously used message-passing algorithm (mpDCA) and can analyze large protein sequences efficiently. The results show that mfDCA accurately predicts a large number of residue contacts, capturing the global structure of the contact map for most protein domains. The method also identifies signals beyond intradomain contacts, such as alternative conformations, ligand-mediated residue couplings, and inter-domain interactions. These findings suggest that DCA-based predictions can be used to guide computational predictions of alternative protein conformations, protein complex formation, and de novo protein domain structure prediction, provided there are sufficient homologous sequences. The study demonstrates that DCA can predict residue contacts with high accuracy, outperforming simple covariance analysis and approximate Bayesian methods. The results also show that DCA can identify long-distance residue pairs that are physically close in structure, indicating that the method is effective for a wide range of protein families, including bacterial and eukaryotic proteins. The study highlights the potential of DCA for sequence-based protein structure prediction and provides insights into the biological significance of residue contacts in protein domains.A computationally efficient implementation of Direct Coupling Analysis (DCA) has been developed to predict residue contacts in protein domains based on sequence information. This method, called mfDCA, is significantly faster than the previously used message-passing algorithm (mpDCA) and can analyze large protein sequences efficiently. The results show that mfDCA accurately predicts a large number of residue contacts, capturing the global structure of the contact map for most protein domains. The method also identifies signals beyond intradomain contacts, such as alternative conformations, ligand-mediated residue couplings, and inter-domain interactions. These findings suggest that DCA-based predictions can be used to guide computational predictions of alternative protein conformations, protein complex formation, and de novo protein domain structure prediction, provided there are sufficient homologous sequences. The study demonstrates that DCA can predict residue contacts with high accuracy, outperforming simple covariance analysis and approximate Bayesian methods. The results also show that DCA can identify long-distance residue pairs that are physically close in structure, indicating that the method is effective for a wide range of protein families, including bacterial and eukaryotic proteins. The study highlights the potential of DCA for sequence-based protein structure prediction and provides insights into the biological significance of residue contacts in protein domains.