Deep learning describes a class of machine learning algorithms that are capable of combining raw inputs into layers of intermediate features. These algorithms have recently shown impressive results across a variety of domains. Biology and medicine are data-rich disciplines, but the data are complex and often ill-understood. Hence, deep learning techniques may be particularly well suited to solve problems of these fields. We examine applications of deep learning to a variety of biomedical problems-patient classification, fundamental biological processes and treatment of patients-and discuss whether deep learning will be able to transform these tasks or if the biomedical sphere poses unique challenges. Following from an extensive literature review, we find that deep learning has yet to revolutionize biomedicine or definitively resolve any of the most pressing challenges in the field, but promising advances have been made on the prior state of the art. Even though improvements over previous baselines have been modest in general, the recent progress indicates that deep learning methods will provide valuable means for speeding up or aiding human investigation. Though progress has been made linking a specific neural network's prediction to input features, understanding how users should interpret these models to make testable hypotheses about the system under study remains an open challenge. Furthermore, the limited amount of labelled data for training presents problems in some domains, as do legal and privacy constraints on work with sensitive health records. Nonetheless, we foresee deep learning enabling changes at both bench and bedside with the potential to transform several areas of biology and medicine.

https://royalsocietypublishing.org/doi/pdf/10.1098/rsif.2017.0387

Opportunities and obstacles for deep learning in biology and medicine.

Single-cell transcriptomics allows researchers to investigate complex communities of heterogeneous cells. It can be applied to stem cells and their descendants in order to chart the progression from multipotent progenitors to fully differentiated cells. While a variety of statistical and computational methods have been proposed for inferring cell lineages, the problem of accurately characterizing multiple branching lineages remains difficult to solve. We introduce Slingshot, a novel method for inferring cell lineages and pseudotimes from single-cell gene expression data. In previously published datasets, Slingshot correctly identifies the biological signal for one to three branching trajectories. Additionally, our simulation study shows that Slingshot infers more accurate pseudotimes than other leading methods. Slingshot is a uniquely robust and flexible tool which combines the highly stable techniques necessary for noisy single-cell data with the ability to identify multiple trajectories. Accurate lineage inference is a critical step in the identification of dynamic temporal gene expression.

/pdf/slingshot-cell-lineage-and-pseudotime-inference-for-single-2xc55fkwkf.pdf

Slingshot: cell lineage and pseudotime inference for single-cell transcriptomics

We have devised and implemented a novel computational strategy for de novo design of molecules with desired properties termed ReLeaSE (Reinforcement Learning for Structural Evolution). On the basis of deep and reinforcement learning (RL) approaches, ReLeaSE integrates two deep neural networks—generative and predictive—that are trained separately but are used jointly to generate novel targeted chemical libraries. ReLeaSE uses simple representation of molecules by their simplified molecular-input line-entry system (SMILES) strings only. Generative models are trained with a stack-augmented memory network to produce chemically feasible SMILES strings, and predictive models are derived to forecast the desired properties of the de novo–generated compounds. In the first phase of the method, generative and predictive models are trained separately with a supervised learning algorithm. In the second phase, both models are trained jointly with the RL approach to bias the generation of new chemical structures toward those with the desired physical and/or biological properties. In the proof-of-concept study, we have used the ReLeaSE method to design chemical libraries with a bias toward structural complexity or toward compounds with maximal, minimal, or specific range of physical properties, such as melting point or hydrophobicity, or toward compounds with inhibitory activity against Janus protein kinase 2. The approach proposed herein can find a general use for generating targeted chemical libraries of novel compounds optimized for either a single desired property or multiple properties.

Deep reinforcement learning for de novo drug design

INTRODUCTION Cancer is one of the leading causes of death worldwide. Although the 2% of the human genome that encodes proteins has been extensively studied, much remains to be learned about the noncoding genome and gene regulation in cancer. Genes are turned on and off in the proper cell types and cell states by transcription factor (TF) proteins acting on DNA regulatory elements that are scattered over the vast noncoding genome and exert long-range influences. The Cancer Genome Atlas (TCGA) is a global consortium that aims to accelerate the understanding of the molecular basis of cancer. TCGA has systematically collected DNA mutation, methylation, RNA expression, and other comprehensive datasets from primary human cancer tissue. TCGA has served as an invaluable resource for the identification of genomic aberrations, altered transcriptional networks, and cancer subtypes. Nonetheless, the gene regulatory landscapes of these tumors have largely been inferred through indirect means. RATIONALE A hallmark of active DNA regulatory elements is chromatin accessibility. Eukaryotic genomes are compacted in chromatin, a complex of DNA and proteins, and only the active regulatory elements are accessible by the cell’s machinery such as TFs. The assay for transposase-accessible chromatin using sequencing (ATAC-seq) quantifies DNA accessibility through the use of transposase enzymes that insert sequencing adapters at these accessible chromatin sites. ATAC-seq enables the genome-wide profiling of TF binding events that orchestrate gene expression programs and give a cell its identity. RESULTS We generated high-quality ATAC-seq data in 410 tumor samples from TCGA, identifying diverse regulatory landscapes across 23 cancer types. These chromatin accessibility profiles identify cancer- and tissue-specific DNA regulatory elements that enable classification of tumor subtypes with newly recognized prognostic importance. We identify distinct TF activities in cancer based on differences in the inferred patterns of TF-DNA interaction and gene expression. Genome-wide correlation of gene expression and chromatin accessibility predicts tens of thousands of putative interactions between distal regulatory elements and gene promoters, including key oncogenes and targets in cancer immunotherapy, such as MYC , SRC , BCL2 , and PDL1 . Moreover, these regulatory interactions inform known genetic risk loci linked to cancer predisposition, nominating biochemical mechanisms and target genes for many cancer-linked genetic variants. Lastly, integration with mutation profiling by whole-genome sequencing identifies cancer-relevant noncoding mutations that are associated with altered gene expression. A single-base mutation located 12 kilobases upstream of the FGD4 gene, a regulator of the actin cytoskeleton, generates a putative de novo binding site for an NKX TF and is associated with an increase in chromatin accessibility and a concomitant increase in FGD4 gene expression. CONCLUSION The accessible genome of primary human cancers provides a wealth of information on the susceptibility, mechanisms, prognosis, and potential therapeutic strategies of diverse cancer types. Prediction of interactions between DNA regulatory elements and gene promoters sets the stage for future integrative gene regulatory network analyses. The discovery of hundreds of noncoding somatic mutations that exhibit allele-specific regulatory effects suggests a pervasive mechanism for cancer cells to manipulate gene expression and increase cellular fitness. These data may serve as a foundational resource for the cancer research community.

The chromatin accessibility landscape of primary human cancers

In the field of artificial intelligence, a combination of scale in data and model capacity enabled by unsupervised learning has led to major advances in representation learning and statistical generation In the life sciences, the anticipated growth of sequencing promises unprecedented data on natural sequence diversity Protein language modeling at the scale of evolution is a logical step toward predictive and generative artificial intelligence for biology To this end, we use unsupervised learning to train a deep contextual language model on 86 billion amino acids across 250 million protein sequences spanning evolutionary diversity The resulting model contains information about biological properties in its representations The representations are learned from sequence data alone The learned representation space has a multiscale organization reflecting structure from the level of biochemical properties of amino acids to remote homology of proteins Information about secondary and tertiary structure is encoded in the representations and can be identified by linear projections Representation learning produces features that generalize across a range of applications, enabling state-of-the-art supervised prediction of mutational effect and secondary structure and improving state-of-the-art features for long-range contact prediction

https://www.pnas.org/content/pnas/118/15/e2016239118.full.pdf

Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences

Local Indicators of Spatial Aggregation (LISA) can be used as objectives in a multicriteria framework when highly autocorrelated areas (hot-spots) must be identified and geographically located in complex areas. To do so, a Multi-Objective Evolutionary Algorithm (MOEA) based on SPEA2 (Strength Pareto Evolutionary Algorithm v.2) has been designed to evaluate three different fitness functions (fine-grained strength, the weighted sum of objectives and fuzzy evaluation of weighted objectives) and three LISA methods. MOEA makes it possible to achieve a compromise between spatial econometric methods as it highlights areas where a specific phenomenon shows significantly high autocorrelation. The spatial distribution of financially compromised olive-tree farms in Andalusia (Spain) was selected for analysis and two fuzzy hot-spots were statistically identified and spatially located. Hot-spots can be considered to be spatial fuzzy sets where the spatial units have a membership degree that can also be calculated.

/pdf/multiobjective-evolutionary-algorithms-to-identify-highly-4g00yfahq1.pdf

Juan C. Fernández-Caballero

Papers

Multiobjective evolutionary algorithms to identify highly autocorrelated areas: the case of spatial distribution in financially compromised farms