An introduction to deep learning on biological sequence data: examples and solutions

In recent years, deep artificial neural networks (including recurrent ones) have won numerous contests in pattern recognition and machine learning. This historical survey compactly summarises relevant work, much of it from the previous millennium. Shallow and deep learners are distinguished by the depth of their credit assignment paths, which are chains of possibly learnable, causal links between actions and effects. I review deep supervised learning (also recapitulating the history of backpropagation), unsupervised learning, reinforcement learning & evolutionary computation, and indirect search for short programs encoding deep and large networks.

The protein-folding problem was first posed about one half-century ago. The term refers to three broad questions: (i) What is the physical code by which an amino acid sequence dictates a protein's native structure? (ii) How can proteins fold so fast? (iii) Can we devise a computer algorithm to predict protein structures from their sequences? We review progress on these problems. In a few cases, computer simulations of the physical forces in chemically detailed models have now achieved the accurate folding of small proteins. We have learned that proteins fold rapidly because random thermal motions cause conformational changes leading energetically downhill toward the native structure, a principle that is captured in funnel-shaped energy landscapes. And thanks in part to the large Protein Data Bank of known structures, predicting protein structures is now far more successful than was thought possible in the early days. What began as three questions of basic science one half-century ago has now grown into the full-fledged research field of protein physical science.

Background The major histocompatibility complex (MHC) molecule plays a central role in controlling the adaptive immune response to infections. MHC class I molecules present peptides derived from intracellular proteins to cytotoxic T cells, whereas MHC class II molecules stimulate cellular and humoral immunity through presentation of extracellularly derived peptides to helper T cells. Identification of which peptides will bind a given MHC molecule is thus of great importance for the understanding of host-pathogen interactions, and large efforts have been placed in developing algorithms capable of predicting this binding event. Results Here, we present a novel artificial neural network-based method, NN-align that allows for simultaneous identification of the MHC class II binding core and binding affinity. NN-align is trained using a novel training algorithm that allows for correction of bias in the training data due to redundant binding core representation. Incorporation of information about the residues flanking the peptide-binding core is shown to significantly improve the prediction accuracy. The method is evaluated on a large-scale benchmark consisting of six independent data sets covering 14 human MHC class II alleles, and is demonstrated to outperform other state-of-the-art MHC class II prediction methods. Conclusion The NN-align method is competitive with the state-of-the-art MHC class II peptide binding prediction algorithms. The method is publicly available at http://www.cbs.dtu.dk/services/NetMHCII-2.0.