The Rosetta software suite for macromolecular modeling, docking, and design is widely used in pharmaceutical, industrial, academic, non-profit, and government laboratories. Despite its broad modeling capabilities, Rosetta remains consistently among leading software suites when compared to other methods created for highly specialized protein modeling and design tasks. Developed for over two decades by a global community of over 60 laboratories, Rosetta has undergone multiple refactorings, and now comprises over three million lines of code. Here we discuss methods developed in the last five years in Rosetta, involving the latest protocols for structure prediction; protein-protein and protein-small molecule docking; protein structure and interface design; loop modeling; the incorporation of various types of experimental data; modeling of peptides, antibodies and proteins in the immune system, nucleic acids, non-standard chemistries, carbohydrates, and membrane proteins. We briefly discuss improvements to the energy function, user interfaces, and usability of the software. Rosetta is available at www.rosettacommons.org.

Flexible peptides that fold upon binding to another protein molecule mediate a large number of regulatory interactions in the living cell and may provide highly specific recognition modules. We present Rosetta FlexPepDock ab-initio, a protocol for simultaneous docking and de-novo folding of peptides, starting from an approximate specification of the peptide binding site. Using the Rosetta fragments library and a coarse-grained structural representation of the peptide and the receptor, FlexPepDock ab-initio samples efficiently and simultaneously the space of possible peptide backbone conformations and rigid-body orientations over the receptor surface of a given binding site. The subsequent all-atom refinement of the coarse-grained models includes full side-chain modeling of both the receptor and the peptide, resulting in high-resolution models in which key side-chain interactions are recapitulated. The protocol was applied to a benchmark in which peptides were modeled over receptors in either their bound backbone conformations or in their free, unbound form. Near-native peptide conformations were identified in 18/26 of the bound cases and 7/14 of the unbound cases. The protocol performs well on peptides from various classes of secondary structures, including coiled peptides with unusual turns and kinks. The results presented here significantly extend the scope of state-of-the-art methods for high-resolution peptide modeling, which can now be applied to a wide variety of peptide-protein interactions where no prior information about the peptide backbone conformation is available, enabling detailed structure-based studies and manipulation of those interactions.

Summary HDAC8 is a member of the family of Histone Deacetylases (HDAC) that catalyze the deacetylation of acetyl lysine residues within histone and non-histone proteins. The recent identification of novel non-histone HDAC8 substrates such as SMC3, ERRα and ARID1A indicates a complex functionality of this enzyme in cellular homeostasis. To discover additional HDAC8 substrates we developed a comprehensive, structure-based approach based on Rosetta FlexPepBind, a protocol that evaluates peptide-binding ability to a receptor from structural models of this interaction. Here we adapt this protocol to identify HDAC8 substrates using peptide sequences extracted from proteins with known acetylated sites. The many new in vitro HDAC8 peptide substrates identified in this study suggest that numerous cellular proteins are HDAC8 substrates, thus expanding our view of the acetylome and its regulation by HDAC8.

Many drugs are developed for commonly occurring, well studied cancer drivers such as vemurafenib for BRAF V600E and erlotinib for EGFR exon 19 mutations. However, most tumors also harbor mutations which have an uncertain role in disease formation, commonly called Variants of Uncertain Significance (VUS), which are not studied or characterized and could play a significant role in drug resistance and relapse. Therefore, the determination of the functional significance of VUS and their response to Molecularly Targeted Agents (MTA) is essential for developing new drugs and predicting response of patients. Here we present a multi-scale deep convolutional neural network (DCNN) architecture combined with an in-vitro functional assay to investigate the functional role of VUS and their response to MTA's. Our method achieved high accuracy and precision on a hold-out set of examples (0.98 mean AUC for all tested genes) and was used to predict the oncogenicity of 195 VUS in 6 genes. 63 (32%) of the assayed VUS's were classified as pathway activating, many of them to a similar extent as known driver mutations. Finally, we show that responses of various mutations to FDA approved MTAs are accurately predicted by our platform in a dose dependent manner. Taken together this novel system can uncover the treatable mutational landscape of a drug and be a useful tool in drug development.

e14615 Background: The MAPK\ERK signaling pathway is a major determinant in the control of diverse cellular processes. This pathway is often up-regulated in human tumors and as such has been an attractive target for the development of anticancer drugs. Mapping the activation landscape of MAPK\ERK is a necessary step in understanding tumor progression and determining the significance of a mutation as well as the relevance of available treatments. Here we present a CNN model, combined with an in-vitro system to investigate the activation landscape of various members of the MAPK\ERK pathway Methods: A novel multi-resolution multi-channel deep neural network was trained to classify images of cells that were transfected with known activating mutated forms of MAPK activating proteins versus WT forms of the proteins. The cells were also transfected with a fluorescently tagged ERK2 expression construct that was one of the three channels imaged by a fluorescent microscopy system. The trained network was subsequently tested on its ability to score images of cells that were transfected with VUS forms of the proteins, or images of cells that were transfected with VUS forms of MAPK proteins and were incubated with the relevant targeted therapies Results: We analyzed 6 proteins activating MAPK\ERK carrying over 100 unique VUS as well as many known oncogenic mutations and determined the pathogenicity score outputted by the network. Analysis of the performance of our model showed that our predictions exceed 95% accuracy and AUC > 0.95. Interestingly, analysis of deeper layers of the network revealed that each gene and mutation exhibit a unique ERK2 activation signal which is considerably different between active and non-active mutations as well as the ability to recognize what mutated gene was transfected, which means that each mutated gene induces different phenological attributes to the cells Conclusions: The use of deep-learning algorithms is providing increasing clinical relevance in several medical fields, most notably pathology. We show that using a cell-based assay combined with high throughput microscopy and CNN based analysis can be used to determine the activity of a wide set of mutations and potentially assist the prediction of targeted therapy outcome. Our results highlight the role of functional interpretation of molecular profiles, enabling more accurate prediction of oncogenic mutations.