DeepNeo: a webserver for predicting immunogenic neoantigens.

Non-self epitopes, whether originated from foreign substances or somatic mutations, trigger immune responses when presented by major histocompatibility complex (MHC) molecules and recognized by T cells. Identification of immunogenically active neoepitopes has significant implications in cancer and virus medicine. However, current methods are mostly limited to predicting physical binding of mutant peptides and MHCs. We previously developed a deep-learning based model, DeepNeo, to identify immunogenic neoepitopes by capturing the structural properties of peptide-MHC pairs with T cell reactivity. Here, we upgraded our DeepNeo model with up-to-date training data. The upgraded model (DeepNeo-v2) was improved in evaluation metrics and showed prediction score distribution that better fits known neoantigen behavior. The immunogenic neoantigen prediction can be conducted at https://deepneo.net.

MHC II immunogenicity shapes the neoepitope landscape in human tumors.

Despite advances in predicting physical peptide-major histocompatibility complex I (pMHC I) binding, it remains challenging to identify functionally immunogenic neoepitopes, especially for MHC II. By using the results of >36,000 immunogenicity assay, we developed a method to identify pMHC whose structural alignment facilitates T cell reaction. Our method predicted neoepitopes for MHC II and MHC I that were responsive to checkpoint blockade when applied to >1,200 samples of various tumor types. To investigate selection by spontaneous immunity at the single epitope level, we analyzed the frequency spectrum of >25 million mutations in >9,000 treatment-naive tumors with >100 immune phenotypes. MHC II immunogenicity specifically lowered variant frequencies in tumors under high immune pressure, particularly with high TCR clonality and MHC II expression. A similar trend was shown for MHC I neoepitopes, but only in particular tissue types. In summary, we report immune selection imposed by MHC II-restricted natural or therapeutic T cell reactivity.

LncRNA HSPA7 in human atherosclerotic plaques sponges miR-223 and promotes the proinflammatory vascular smooth muscle cell transition.

Although there are many genetic loci in noncoding regions associated with vascular disease, studies on long noncoding RNAs (lncRNAs) discovered from human plaques that affect atherosclerosis have been highly limited. We aimed to identify and functionally validate a lncRNA using human atherosclerotic plaques. Human aortic samples were obtained from patients who underwent aortic surgery, and tissues were classified according to atherosclerotic plaques. RNA was extracted and analyzed for differentially expressed lncRNAs in plaques. Human aortic smooth muscle cells (HASMCs) were stimulated with oxidized low-density lipoprotein (oxLDL) to evaluate the effect of the identified lncRNA on the inflammatory transition of the cells. Among 380 RNAs differentially expressed between the plaque and control tissues, lncRNA HSPA7 was selected and confirmed to show upregulated expression upon oxLDL treatment. HSPA7 knockdown inhibited the migration of HASMCs and the secretion and expression of IL-1ß and IL-6; however, HSPA7 knockdown recovered the oxLDL-induced reduction in the expression of contractile markers. Although miR-223 inhibition promoted the activity of Nf-κB and the secretion of inflammatory proteins such as IL-1ß and IL-6, HSPA7 knockdown diminished these effects. The effects of miR-223 inhibition and HSPA7 knockdown were also found in THP-1 cell-derived macrophages. The impact of HSPA7 on miR-223 was mediated in an AGO2-dependent manner. HSPA7 is differentially increased in human atheroma and promotes the inflammatory transition of vascular smooth muscle cells by sponging miR-223. For the first time, this study elucidated the molecular mechanism of action of HSPA7, a lncRNA of previously unknown function, in humans.

Accelerated Evolution of the Regulatory Sequences of Brain Development in the Human Genome.

Genetic modifications in noncoding regulatory regions are likely critical to human evolution. Human-accelerated noncoding elements are highly conserved noncoding regions among vertebrates but have large differences across humans, which implies human-specific regulatory potential. In this study, we found that human-accelerated noncoding elements were frequently coupled with DNase I hypersensitive sites (DHSs), together with monomethylated and trimethylated histone H3 lysine 4, which are active regulatory markers. This coupling was particularly pronounced in fetal brains relative to adult brains, non-brain fetal tissues, and embryonic stem cells. However, fetal brain DHSs were also specifically enriched in deeply conserved sequences, implying coexistence of universal maintenance and human-specific fitness in human brain development. We assessed whether this coexisting pattern was a general one by quantitatively measuring evolutionary rates of DHSs. As a result, fetal brain DHSs showed a mixed but distinct signature of regional conservation and outlier point acceleration as compared to other DHSs. This finding suggests that brain developmental sequences are selectively constrained in general, whereas specific nucleotides are under positive selection or constraint relaxation simultaneously. Hence, we hypothesize that human- or primate-specific changes to universally conserved regulatory codes of brain development may drive the accelerated, and most likely adaptive, evolution of the regulatory network of the human brain.

Predictive long-range allele-specific mapping of regulatory variants and target transcripts.

Genome-wide association studies (GWASs) have identified a large number of noncoding associations, calling for systematic mapping to causal regulatory variants and their distal target genes. A widely used method, quantitative trait loci (QTL) mapping for chromatin or expression traits, suffers from sample-to-sample experimental variation and trans-acting or environmental effects. Instead, alleles at heterozygous loci can be compared within a sample, thereby controlling for those confounding factors. Here we introduce a method for chromatin structure-based allele-specific pairing of regulatory variants and target transcripts. With phased genotypes, much of allele-specific expression could be explained by paired allelic cis-regulation across a long range. This approach showed approximately two times greater sensitivity than QTL mapping. There are cases in which allele imbalance cannot be tested because heterozygotes are not available among reference samples. Therefore, we employed a machine learning method to predict missing positive cases based on various features shared by observed allele-specific pairs. We showed that only 10 reference samples are sufficient to achieve high prediction accuracy with a low sampling variation. In conclusion, our method enables highly sensitive fine mapping and target identification for trait-associated variants based on a small number of reference samples.

Predicting the recurrence of noncoding regulatory mutations in cancer.

BACKGROUND: One of the greatest challenges in cancer genomics is to distinguish driver mutations from passenger mutations. Whereas recurrence is a hallmark of driver mutations, it is difficult to observe recurring noncoding mutations owing to a limited amount of whole-genome sequenced samples. Hence, it is required to develop a method to predict potentially recurrent mutations. RESULTS: In this work, we developed a random forest classifier that predicts regulatory mutations that may recur based on the features of the mutations repeatedly appearing in a given cohort. With breast cancer as a model, we profiled 35 quantitative features describing genetic and epigenetic signals at the mutation site, transcription factors whose binding motif was disrupted by the mutation, and genes targeted by long-range chromatin interactions. A true set of mutations for machine learning was generated by interrogating publicly available pan-cancer genomes based on our statistical model of mutation recurrence. The performance of our random forest classifier was evaluated by cross validations. The variable importance of each feature in the classification of mutations was investigated. Our statistical recurrence model for the random forest classifier showed an area under the curve (AUC) of ~0.78 in predicting recurrent mutations. Chromatin accessibility at the mutation sites, the distance from the mutations to known cancer risk loci, and the role of the target genes in the regulatory or protein interaction network were among the most important variables. CONCLUSIONS: Our methods enable to characterize recurrent regulatory mutations using a limited number of whole-genome samples, and based on the characterization, to predict potential driver mutations whose recurrence is not found in the given samples but likely to be observed with additional samples.

Intersection of genetics and epigenetics in monozygotic twin genomes.

As a final function of various epigenetic mechanisms, chromatin regulation is a transcription control process that especially demonstrates active interaction with genetic elements. Thus, chromatin structure has become a principal focus in recent genomics researches that strive to characterize regulatory functions of DNA variants related to diseases or other traits. Although researchers have been focusing on DNA methylation when studying monozygotic (MZ) twins, a great model in epigenetics research, interactions between genetics and epigenetics in chromatin level are expected to be an imperative research trend in the future. In this review, we discuss how the genome, epigenome, and transcriptome of MZ twins can be studied in an integrative manner from this perspective.

Genetic Architecture of Transcription and Chromatin Regulation.

DNA microarray and next-generation sequencing provide data that can be used for the genetic analysis of multiple quantitative traits such as gene expression levels, transcription factor binding profiles, and epigenetic signatures. In particular, chromatin opening is tightly coupled with gene transcription. To understand how these two processes are genetically regulated and associated with each other, we examined the changes of chromatin accessibility and gene expression in response to genetic variation by means of quantitative trait loci mapping. Regulatory patterns commonly observed in yeast and human across different technical platforms and experimental designs suggest a higher genetic complexity of transcription regulation in contrast to a more robust genetic architecture of chromatin regulation.

Global transcription network incorporating distal regulator binding reveals selective cooperation of cancer drivers and risk genes.

Global network modeling of distal regulatory interactions is essential in understanding the overall architecture of gene expression programs. Here, we developed a Bayesian probabilistic model and computational method for global causal network construction with breast cancer as a model. Whereas physical regulator binding was well supported by gene expression causality in general, distal elements in intragenic regions or loci distant from the target gene exhibited particularly strong functional effects. Modeling the action of long-range enhancers was critical in recovering true biological interactions with increased coverage and specificity overall and unraveling regulatory complexity underlying tumor subclasses and drug responses in particular. Transcriptional cancer drivers and risk genes were discovered based on the network analysis of somatic and genetic cancer-related DNA variants. Notably, we observed that the risk genes were functionally downstream of the cancer drivers and were selectively susceptible to network perturbation by tumorigenic changes in their upstream drivers. Furthermore, cancer risk alleles tended to increase the susceptibility of the transcription of their associated genes. These findings suggest that transcriptional cancer drivers selectively induce a combinatorial misregulation of downstream risk genes, and that genetic risk factors, mostly residing in distal regulatory regions, increase transcriptional susceptibility to upstream cancer-driving somatic changes.

Understanding Epistatic Interactions between Genes Targeted by Non-coding Regulatory Elements in Complex Diseases.

Genome-wide association studies have proven the highly polygenic architecture of complex diseases or traits; therefore, single-locus-based methods are usually unable to detect all involved loci, especially when individual loci exert small effects. Moreover, the majority of associated single-nucleotide polymorphisms resides in non-coding regions, making it difficult to understand their phenotypic contribution. In this work, we studied epistatic interactions associated with three common diseases using Korea Association Resource (KARE) data: type 2 diabetes mellitus (DM), hypertension (HT), and coronary artery disease (CAD). We showed that epistatic single-nucleotide polymorphisms (SNPs) were enriched in enhancers, as well as in DNase I footprints (the Encyclopedia of DNA Elements [ENCODE] Project Consortium 2012), which suggested that the disruption of the regulatory regions where transcription factors bind may be involved in the disease mechanism. Accordingly, to identify the genes affected by the SNPs, we employed whole-genome multiple-cell-type enhancer data which discovered using DNase I profiles and Cap Analysis Gene Expression (CAGE). Assigned genes were significantly enriched in known disease associated gene sets, which were explored based on the literature, suggesting that this approach is useful for detecting relevant affected genes. In our knowledge-based epistatic network, the three diseases share many associated genes and are also closely related with each other through many epistatic interactions. These findings elucidate the genetic basis of the close relationship between DM, HT, and CAD.