Advances in biological and medical technologies have been providing us explosive volumes of biological and physiological data, such as medical images, electroencephalography, genomic and protein sequences. Learning from these data facilitates the understanding of human health and disease. Developed from artificial neural networks, deep learning-based algorithms show great promise in extracting features and learning patterns from complex data. The aim of this paper is to provide an overview of deep learning techniques and some of the state-of-the-art applications in the biomedical field. We first introduce the development of artificial neural network and deep learning. We then describe two main components of deep learning, i.e., deep learning architectures and model optimization. Subsequently, some examples are demonstrated for deep learning applications, including medical image classification, genomic sequence analysis, as well as protein structure classification and prediction. Finally, we offer our perspectives for the future directions in the field of deep learning.

DNA methylation is an important type of epigenetic modifications, where 5- methylcytosine (5mC), 6-methyadenine (6mA) and 4-methylcytosine (4mC) are the most common types. Previous efforts have been largely focused on 5mC, providing invaluable insights into epigenetic regulation through DNA methylation. Recently developed single-molecule real-time (SMRT) sequencing technology provides a unique opportunity to detect the less studied DNA 6mA and 4mC modifications at single-nucleotide resolution. With a rapidly increased amount of SMRT sequencing data generated, there is an emerging demand to systematically explore DNA 6mA and 4mC modifications from these data sets. MethSMRT is the first resource hosting DNA 6mA and 4mC methylomes. All the data sets were processed using the same analysis pipeline with the same quality control. The current version of the database provides a platform to store, browse, search and download epigenome-wide methylation profiles of 156 species, including seven eukaryotes such as Arabidopsis, C. elegans, Drosophila, mouse and yeast, as well as 149 prokaryotes. It also offers a genome browser to visualize the methylation sites and related information such as single nucleotide polymorphisms (SNP) and genomic annotation. Furthermore, the database provides a quick summary of statistics of methylome of 6mA and 4mC and predicted methylation motifs for each species. MethSMRT is publicly available at http://sysbio.sysu.edu.cn/methsmrt/ without use restriction.

DNA variants represent an important source of genetic variations among individuals. Next- generation sequencing (NGS) is the most popular technology for genome-wide variant calling. Third-generation sequencing (TGS) has also recently been used in genetic studies. Although many variant callers are available, no single caller can call both types of variants on NGS or TGS data with high sensitivity and specificity. In this study, we systematically evaluated 11 variant callers on 12 NGS and TGS datasets. For germline variant calling, we tested DNAseq and DNAscope modes from Sentieon, HaplotypeCaller mode from GATK and WGS mode from DeepVariant. All the four callers had comparable performance on NGS data and 30× coverage of WGS data was recommended. For germline variant calling on TGS data, we tested DNAseq mode from Sentieon, HaplotypeCaller mode from GATK and PACBIO mode from DeepVariant. All the three callers had similar performance in SNP calling, while DeepVariant outperformed the others in InDel calling. TGS detected more variants than NGS, particularly in complex and repetitive regions. For somatic variant calling on NGS, we tested TNscope and TNseq modes from Sentieon, MuTect2 mode from GATK, NeuSomatic, VarScan2, and Strelka2. TNscope and Mutect2 outperformed the other callers. A higher proportion of tumor sample purity (from 10 to 20%) significantly increased the recall value of calling. Finally, computational costs of the callers were compared and Sentieon required the least computational cost. These results suggest that careful selection of a tool and parameters is needed for accurate SNP or InDel calling under different scenarios.

Glioblastoma multiforme (GBM) is among the most aggressive of human cancers. Although differentiation therapy has been proposed as a potential approach to treat GBM, the mechanisms of induced differentiation remain poorly defined. Here, we established an induced differentiation model of GBM using cAMP activators that specifically directed GBM differentiation into astroglia. Transcriptomic and proteomic analyses revealed that oxidative phosphorylation and mitochondrial biogenesis are involved in induced differentiation of GBM. Dibutyryl cyclic AMP (dbcAMP) reverses the Warburg effect, as evidenced by increased oxygen consumption and reduced lactate production. Mitochondrial biogenesis induced by activation of the CREB-PGC1α pathway triggers metabolic shift and differentiation. Blocking mitochondrial biogenesis using mdivi1 or by silencing PGC1α abrogates differentiation; conversely, overexpression of PGC1α elicits differentiation. In GBM xenograft models and patient-derived GBM samples, cAMP activators also induce tumor growth inhibition and differentiation. Our data show that mitochondrial biogenesis and metabolic switch to oxidative phosphorylation drive the differentiation of tumor cells.

Human ribosomes have long been thought to be uniform factories with little regulatory function. Accumulating evidence emphasizes the heterogeneity of ribosomal protein (RP) expression in specific cellular functions and development. However, a systematic understanding of functional relevance of RPs is lacking. Here, we surveyed translational and transcriptional changes after individual knockdown of 75 RPs, 44 from the large subunit (60S) and 31 from the small subunit (40S), by Ribo-seq and RNA-seq analyses. Deficiency of individual RPs altered specific subsets of genes transcriptionally and translationally. RP genes were under cotranslational regulation upon ribosomal stress, and deficiency of the 60S RPs and the 40S RPs had opposite effects. RP deficiency altered the expression of genes related to eight major functional classes, including the cell cycle, cellular metabolism, signal transduction and development. 60S RP deficiency led to greater inhibitory effects on cell growth than did 40S RP deficiency, through P53 signaling. Particularly, we showed that eS8/RPS8 deficiency stimulated apoptosis while eL13/RPL13 or eL18/RPL18 deficiency promoted senescence. We also validated the phenotypic impacts of uL5/RPL11 and eL15/RPL15 deficiency on retina development and angiogenesis, respectively. Overall, our study provides a valuable resource for and novel insights into ribosome regulation in cellular activities, development and diseases. Ribosomes are the main effector of the translational machinery to synthesize proteins. In this study, the authors characterized genome-wide transcriptional and translational changes after knocking-down 75 individual human ribosomal proteins (RPs). They revealed that deficiency of individual RPs perturbed expression of specific subsets of genes, enriched in eight major functional classes, such as cell cycle and development. RPs were subjected to co-translational regulation under ribosomal stress where deficiency of the 60S RPs and the 40S RPs had opposite effects on the two subunits. They also showed that RPS8 deficiency stimulated cellular apoptosis while RPL13 and RPL18 deficiency promoted cellular senescence. They further showed functional and regulatory roles of RPL11 and RPL15 in retina development and angiogenesis, respectively.