Deep Learning (DL) has recently enabled unprecedented advances in one of the grand challenges in computational biology: the half-century-old problem of protein structure prediction. In this paper we discuss recent advances, limitations, and future perspectives of DL on five broad areas: protein structure prediction, protein function prediction, genome engineering, systems biology and data integration, and phylogenetic inference. We discuss each application area and cover the main bottlenecks of DL approaches, such as training data, problem scope, and the ability to leverage existing DL architectures in new contexts. To conclude, we provide a summary of the subject-specific and general challenges for DL across the biosciences.

Understanding the challenges of protein flexibility in drug design. . ???aop.label???. doi: 10.1517/17460441.2015.1094458

Molecular docking is a standard computational approach to predict binding modes of protein–ligand complexes by exploring alternative orientations and conformations of the ligand (i.e., by exploring ligand flexibility). Docking tools are largely used for virtual screening of small drug-like molecules, but their accuracy and efficiency greatly decays for ligands with more than 10 flexible bonds. This prevents a broader use of these tools to dock larger ligands, such as peptides, which are molecules of growing interest in cancer research. To overcome this limitation, our group has previously proposed a meta-docking strategy, called DINC, to predict binding modes of large ligands. By incrementally docking overlapping fragments of a ligand, DINC allowed predicting binding modes of peptide-based inhibitors of transcription factors involved in cancer. Here, we describe DINC 2.0, a revamped version of the DINC webserver with enhanced capabilities and a more user-friendly interface. DINC 2.0 allows docking ligands that were previously too challenging for DINC, such as peptides with more than 25 flexible bonds. The webserver is freely accessible at http://dinc.kavrakilab.org, together with additional documentation and video tutorials. Our team will provide continuous support for this tool and is working on extending its applicability to other challenging fields, such as personalized immunotherapy against cancer. Cancer Res; 77(21); e55–57. ©2017 AACR.

Understanding the mechanisms involved in the activation of an immune response is essential to many fields in human health, including vaccine development and personalized cancer immunotherapy. A central step in the activation of the adaptive immune response is the recognition, by T-cell lymphocytes, of peptides displayed by a special type of receptor known as Major Histocompatibility Complex (MHC). Considering the key role of MHC receptors in T-cell activation, the computational prediction of peptide binding to MHC has been an important goal for many immunological applications. Sequence- based methods have become the gold standard for peptide-MHC binding affinity prediction, but structure-based methods are expected to provide more general predictions (i.e., predictions applicable to all types of MHC receptors). In addition, structural modeling of peptide-MHC complexes has the potential to uncover yet unknown drivers of T-cell activation, thus allowing for the development of better and safer therapies. In this review, we discuss the use of computational methods for the structural modeling of peptide-MHC complexes (i.e., binding mode prediction) and for the structure-based prediction of binding affinity.

Immunotherapy has become one of the most promising avenues for cancer treatment, making use of the patient’s own immune system to eliminate cancer cells. Clinical trials with T-cell-based immunotherapies have shown dramatic tumor regressions, being effective in multiple cancer types and for many different patients. Unfortunately, this progress was tempered by reports of serious (even fatal) side effects. Such therapies rely on the use of cytotoxic T-cell lymphocytes, an essential part of the adaptive immune system. Cytotoxic T-cells are regularly involved in surveillance and are capable of both eliminating diseased cells and generating protective immunological memory. The specificity of a given T-cell is determined through the structural interaction between the T-cell receptor (TCR) and a peptide-loaded major histocompatibility complex (MHC); i.e., an intracellular peptide-ligand displayed at the cell surface by an MHC receptor. However, a given TCR can recognize different peptide-MHC (pMHC) complexes, which can sometimes trigger an unwanted response that is referred to as T-cell cross-reactivity. This has become a major safety issue in T-cell-based immunotherapies, following reports of melanoma-specific T-cells causing cytotoxic damage to healthy tissues (e.g., heart and nervous system). T-cell cross-reactivity has been extensively studied in the context of viral immunology and tissue transplantation. Growing evidence suggests that it is largely driven by structural similarities of seemingly unrelated pMHC complexes. Here we review recent reports about the existence of pMHC "hot-spots" for cross-reactivity and propose the existence of a TCR interaction profile (i.e., a refinement of a more general TCR footprint in which some amino acid residues are more important than others in triggering T-cell cross-reactivity). We also make use of available structural data and pMHC models to interpret previously reported cross-reactivity patterns among virus-derived peptides. Our study provides further evidence that structural analyses of pMHC complexes can be used to assess the intrinsic likelihood of cross-reactivity among peptide-targets. Furthermore, we hypothesize that some apparent inconsistencies in reported cross-reactivities, such as a preferential directionality, might also be driven by particular structural features of the targeted pMHC complex. Finally, we explain why T-cell-based immunotherapy provides a special context in which meaningful T-cell cross-reactivity predictions can be made

The Class I Major Histocompatibility Complex (MHC) is a central protein in immunology as it binds to intracellular peptides and displays them at the cell surface for recognition by T-cells. The structural analysis of bound peptide-MHC complexes (pMHCs) holds the promise of interpretable and general binding prediction (i.e., testing whether a given peptide binds to a given MHC). However, structural analysis is limited in part by the difficulty in modelling pMHCs given the size and flexibility of the peptides that can be presented by MHCs. This article describes APE-Gen (Anchored Peptide-MHC Ensemble Generator), a fast method for generating ensembles of bound pMHC conformations. APE-Gen generates an ensemble of bound conformations by iterated rounds of (i) anchoring the ends of a given peptide near known pockets in the binding site of the MHC, (ii) sampling peptide backbone conformations with loop modelling, and then (iii) performing energy minimization to fix steric clashes, accumulating conformations at each round. APE-Gen takes only minutes on a standard desktop to generate tens of bound conformations, and we show the ability of APE-Gen to sample conformations found in X-ray crystallography even when only sequence information is used as input. APE-Gen has the potential to be useful for its scalability (i.e., modelling thousands of pMHCs or even non-canonical longer peptides) and for its use as a flexible search tool. We demonstrate an example for studying cross-reactivity.