Two-Way Horizontal and Vertical Omics Integration for Disease Subtype Discovery

Disease subtype discovery is an essential step in delivering personalized medicine. Disease subtyping via omics data has become a common approach for this purpose. With the advancement of technology and the lower price for generating omics data, multi-level and multi-cohort omics data are prevalent in the public domain, providing unprecedented opportunities to decrypt disease mechanisms. How to fully utilize multi-level/multi-cohort omics data and incorporate established biological knowledge toward disease subtyping remains a challenging problem. In this paper, we propose a meta-analytic integrative sparse Kmeans (MISKmeans) algorithm for integrating multi-cohort/multi-level omics data and prior biological knowledge. Compared with previous methods, MISKmeans shows better clustering accuracy and feature selection relevancy. An efficient R package, “MIS-Kmeans”, calling C++ is freely available on GitHub (https://github.com/Caleb-Huo/MIS-Kmeans).

     

CpG methylation of DNA restricts prereplication complex assembly in Xenopus egg extracts

In a Xenopus egg replication system, the origin recognition complex (ORC) does not bind to CpG methylated DNA and DNA replication is inhibited. Insertion of low density CpG DNA of at least 1.2 kb into methylated plasmids rescues both replication and ORC binding. Using this pseudo-origin, we find that ORC binding is restricted to low-CpG-density DNA; however, MCM is loaded onto both weakly and highly methylated DNA and occupies at least approximately 2 kb of DNA. Replication initiates coincident with MCM, and even the most distally bound MCM is associated with sites of replication initiation. These results suggest that in metazoans MCM is loaded onto and initiates replication over a large region distant from ORC.     

COHCAP: an integrative genomic pipeline for single-nucleotide resolution DNA methylation analysis

COHCAP (City of Hope CpG Island Analysis Pipeline) is an algorithm to analyze single-nucleotide resolution DNA methylation data produced by either an Illumina methylation array or targeted bisulfite sequencing. The goal of the COHCAP algorithm is to identify CpG islands that show a consistent pattern of methylation among CpG sites. COHCAP is currently the only DNA methylation package that provides integration with gene expression data to identify a subset of CpG islands that are most likely to regulate downstream gene expression, and it can generate lists of differentially methylated CpG islands with ∼50% concordance with gene expression from both cell line data and heterogeneous patient data. For example, this article describes known breast cancer biomarkers (such as estrogen receptor) with a negative correlation between DNA methylation and gene expression. COHCAP also provides visualization for quality control metrics, regions of differential methylation and correlation between methylation and gene expression. This software is freely available at https://sourceforge.net/projects/cohcap/.     

A powerful method to integrate genotype and gene expression data for dissecting the genetic architecture of a disease

To decipher the genetic architecture of human disease, various types of omics data are generated. Two common omics data are genotypes and gene expression. Often genotype data for a large number of individuals and gene expression data for a few individuals are generated due to biological and technical reasons, leading to unequal sample sizes for different omics data. Unavailability of standard statistical procedure for integrating such datasets motivates us to propose a two-step multi-locus association method using latent variables. Our method is powerful than single/separate omics data analysis and it unravels comprehensively deep-seated signals through a single statistical model. Extensive simulation confirms that it is robust to various genetic models as its power increases with sample size and number of associated loci. It provides p-values very fast. Application to real dataset on psoriasis identifies 17 novel SNPs, functionally related to psoriasis-associated genes, at much smaller sample size than standard GWAS.     

Strong and weak male mutation bias at different sites in the primate genomes: insights from the human-chimpanzee comparison

Male mutation bias is a higher mutation rate in males than in females thought to result from the greater number of germ line cell divisions in males. If errors in DNA replication cause most mutations, then the magnitude of male mutation bias, measured as the male-to-female mutation rate ratio (alpha), should reflect the relative excess of male versus female germ line cell divisions. Evolutionary rates averaged among all sites in a sequence and compared between mammalian sex chromosomes were shown to be indeed higher in males than in females. However, it is presently unknown whether individual classes of substitutions exhibit such bias. To address this issue, we investigated male mutation bias separately at non-CpG and CpG sites using human-chimpanzee whole-genome alignments. We observed strong male mutation bias at non-CpG sites: alpha in the X-autosome comparison was approximately 6-7, which was similar to the male-to-female ratio in the number of germ line cell divisions. In contrast, mutations at CpG sites exhibited weak male mutation bias: alpha in the X-autosome comparison was only approximately 2-3. This is consistent with the methylation-induced and replication-independent mechanism of CpG transitions, which constitute the majority of mutations at CpG sites. Interestingly, our study also indicated weak male mutation bias for transversions at CpG sites, implying a spontaneous mechanism largely not associated with replication. Male mutation bias was equally strong at CpG and non-CpG sites located within unmethylated "CpG islands," suggesting the replication-dependent origin of these mutations. Thus, we found that the strength of male mutation bias is nonuniform in the primate genomes. Importantly, we discovered that male mutation bias depends on the proportion of CpG sites in the loci compared. This might explain the differences in the magnitude of primate male mutation bias observed among studies.     

Improved reduced representation bisulfite sequencing for epigenomic profiling of clinical samples

Background

DNA methylation plays crucial roles in epigenetic gene regulation in normal development and disease pathogenesis. Efficient and accurate quantification of DNA methylation at single base resolution can greatly advance the knowledge of disease mechanisms and be used to identify potential biomarkers. We developed an improved pipeline based on reduced representation bisulfite sequencing (RRBS) for cost-effective genome-wide quantification of DNA methylation at single base resolution. A selection of two restriction enzymes (Taq^αI and MspI) enables a more unbiased coverage of genomic regions of different CpG densities. We further developed a highly automated software package to analyze bisulfite sequencing results from the Solexa GAIIx system.

Results

With two sequencing lanes, we were able to quantify ~1.8 million individual CpG sites at a minimum sequencing depth of 10. Overall, about 76.7% of CpG islands, 54.9% of CpG island shores and 52.2% of core promoters in the human genome were covered with at least 3 CpG sites per region.

Conclusions

With this new pipeline, it is now possible to perform whole-genome DNA methylation analysis at single base resolution for a large number of samples for understanding how DNA methylation and its changes are involved in development, differentiation, and disease pathogenesis.     

TLR Ligands Induce Antiviral Responses in Chicken Macrophages

Chicken macrophages express several receptors for recognition of pathogens, including Toll-like receptors (TLRs). TLRs bind to pathogen-associated molecular patterns (PAMPs) derived from bacterial or viral pathogens leading to the activation of macrophages. Macrophages play a critical role in immunity against viruses, including influenza viruses. The present study was designed to test the hypothesis that treatment of chicken macrophages with TLR ligands reduces avian influenza replication. Furthermore, we sought to study the expression of some of the key mediators involved in the TLR-mediated antiviral responses of macrophages. Chicken macrophages were treated with the TLR2, 3, 4, 7 and 21 ligands, Pam3CSK4, poly(I:C), LPS, R848 and CpG ODN, respectively, at different doses and time points pre- and post-H4N6 avian influenza virus (AIV) infection. The results revealed that pre-treatment of macrophages with Pam3CSK4, LPS and CpG ODN reduced the replication of AIV in chicken macrophages. In addition, the relative expression of genes involved in inflammatory and antiviral responses were quantified at 3, 8 and 18 hours post-treatment with the TLR2, 4 and 21 ligands. Pam3CSK4, LPS and CpG ODN increased the expression of interleukin (IL)-1β, interferon (IFN)-γ, IFN-β and interferon regulatory factor (IFR) 7. The expression of these genes correlated with the reduction of viral replication in macrophages. These results shed light on the process of immunity to AIV in chickens.     

The dynamics of genome replication using deep sequencing

Eukaryotic genomes are replicated from multiple DNA replication origins. We present complementary deep sequencing approaches to measure origin location and activity in Saccharomyces cerevisiae. Measuring the increase in DNA copy number during a synchronous S-phase allowed the precise determination of genome replication. To map origin locations, replication forks were stalled close to their initiation sites; therefore, copy number enrichment was limited to origins. Replication timing profiles were generated from asynchronous cultures using fluorescence-activated cell sorting. Applying this technique we show that the replication profiles of haploid and diploid cells are indistinguishable, indicating that both cell types use the same cohort of origins with the same activities. Finally, increasing sequencing depth allowed the direct measure of replication dynamics from an exponentially growing culture. This is the first time this approach, called marker frequency analysis, has been successfully applied to a eukaryote. These data provide a high-resolution resource and methodological framework for studying genome biology.     

Methylation-mediated silencing of TMS1/ASC is accompanied by histone hypoacetylation and CpG island-localized changes in chromatin architecture.

Aberrant methylation of CpG-dense islands in the promoter regions of genes is an acquired epigenetic alteration associated with the silencing of tumor suppressor genes in human cancers. In a screen for endogenous targets of methylation-mediated gene silencing, we identified a novel CpG island-associated gene, TMS1, which is aberrantly methylated and silenced in response to the ectopic expression of DNA methyltransferase-1. TMS1 functions in the regulation of apoptosis and is frequently methylated and silenced in human breast cancers. In this study, we characterized the methylation pattern and chromatin architecture of the TMS1 locus in normal fibroblasts and determined the changes associated with its progressive methylation. In normal fibroblasts expressing TMS1, the CpG island is defined by an unmethylated domain that is separated from densely methylated flanking DNA by distinct 5' and 3' boundaries. Analysis of the nucleoprotein architecture of the locus in intact nuclei revealed three DNase I-hypersensitive sites that map within the CpG island. Strikingly, two of these sites coincided with the 5'- and 3'-methylation boundaries. Methylation of the TMS1 CpG island was accompanied by loss of hypersensitive site formation, hypoacetylation of histones H3 and H4, and gene silencing. This altered chromatin structure was confined to the CpG island and occurred without significant changes in methylation, histone acetylation, or hypersensitive site formation at a fourth DNase I-hypersensitive site 2 kb downstream of the TMS1 CpG island. The data indicate that there are sites of protein binding and/or structural transitions that define the boundaries of the unmethylated CpG island in normal cells and that aberrant methylation overcomes these boundaries to direct a local change in chromatin structure, resulting in gene silencing.     

Endogenous IFN-alpha production by plasmacytoid dendritic cells exerts an antiviral effect on thymic HIV-1 infection

Plasmacytoid dendritic cells (pDC) are the principal producers of IFN-alpha in response to viral infection. Because pDC are present in the thymus, we investigated the consequences of HIV-1-induced IFN-alpha production by thymic pDC. We observed that thymic pDC as well as thymocytes express intracellular IFN-alpha upon infection with HIV-1. However, only the pDC could suppress HIV-1 replication, because depletion of pDC resulted in enhancement of HIV-1 replication in thymocytes. Thymic pDC could also produce IFN-alpha in response to CpG oligonucleotides, consistent with the observations of others that peripheral pDC produce IFN-alpha upon engagement of TLR-9. Importantly, CpG considerably increased IFN-alpha production induced by HIV-1, and addition of CpG during HIV-1 infection enhanced expression of the IFN response protein MxA in thymocytes and strongly reduced HIV-replication. Our data indicate that thymic pDC modulate HIV-1 replication through secretion of IFN-alpha. The degree of inhibition depends on the level of IFN-alpha produced by the thymic pDC.     

Pheno2Geno - High-throughput generation of genetic markers and maps from molecular phenotypes for crosses between inbred strains

Background

Genetic markers and maps are instrumental in quantitative trait locus (QTL) mapping in segregating populations. The resolution of QTL localization depends on the number of informative recombinations in the population and how well they are tagged by markers. Larger populations and denser marker maps are better for detecting and locating QTLs. Marker maps that are initially too sparse can be saturated or derived de novo from high-throughput omics data, (e.g. gene expression, protein or metabolite abundance). If these molecular phenotypes are affected by genetic variation due to a major QTL they will show a clear multimodal distribution. Using this information, phenotypes can be converted into genetic markers.

Results

The Pheno2Geno tool uses mixture modeling to select phenotypes and transform them into genetic markers suitable for construction and/or saturation of a genetic map. Pheno2Geno excludes candidate genetic markers that show evidence for multiple possibly epistatically interacting QTL and/or interaction with the environment, in order to provide a set of robust markers for follow-up QTL mapping.We demonstrate the use of Pheno2Geno on gene expression data of 370,000 probes in 148 A. thaliana recombinant inbred lines. Pheno2Geno is able to saturate the existing genetic map, decreasing the average distance between markers from 7.1 cM to 0.89 cM, close to the theoretical limit of 0.68 cM (with 148 individuals we expect a recombination every 100/148=0.68 cM); this pinpointed almost all of the informative recombinations in the population.

Conclusion

The Pheno2Geno package makes use of genome-wide molecular profiling and provides a tool for high-throughput de novo map construction and saturation of existing genetic maps. Processing of the showcase dataset takes less than 30 minutes on an average desktop PC. Pheno2Geno improves QTL mapping results at no additional laboratory cost and with minimum computational effort. Its results are formatted for direct use in R/qtl, the leading R package for QTL studies. Pheno2Geno is freely available on CRAN under “GNU GPL v3”. The Pheno2Geno package as well as the tutorial can also be found at: http://pheno2geno.nl.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0475-6) contains supplementary material, which is available to authorized users.     

Promoter hypomethylation contributes to the expression of MUC3A in cancer cells

MUC3A is a membrane-bound glycoprotein that is aberrantly expressed in carcinomas and is a risk factor for a poor prognosis. However, the exact mechanism of MUC3A expression has yet to be clarified. Here, we provide the first evidence that MUC3A gene expression is controlled by the CpG methylation status of the proximal promoter region. We show that the DNA methylation pattern is intimately correlated with MUC3A expression in breast, lung, pancreas and colon cancer cell lines. The DNA methylation status of 30 CpG sites from −660 to +273 was mapped using MassARRAY analysis. MUC3A-negative cancer cell lines and those with low MUC3A expression (e.g., MCF-7) were highly methylated in the proximal promoter region, corresponding to 9 CpG sites (−345 to −75 bp), whereas MUC3A-positive cell lines (e.g., LS174T) had low methylation levels. Moreover, 5-aza-2′-deoxycytidine and trichostatin A treatment of MUC3A-negative cells or those with low MUC3A expression caused elevation of MUC3A mRNA. Our results suggest that DNA hypomethylation in the 5′-flanking region of the MUC3A gene plays an important role in MUC3A expression in carcinomas of various organs. An understanding of epigenetic changes in MUC3A may contribute to the diagnosis of carcinogenic risk and to prediction of outcome in patients with cancer.     

AIME: Autoencoder-based integrative multi-omics data embedding that allows for confounder adjustments

In the integrative analyses of omics data, it is often of interest to extract data representation from one data type that best reflect its relations with another data type. This task is traditionally fulfilled by linear methods such as canonical correlation analysis (CCA) and partial least squares (PLS). However, information contained in one data type pertaining to the other data type may be complex and in nonlinear form. Deep learning provides a convenient alternative to extract low-dimensional nonlinear data embedding. In addition, the deep learning setup can naturally incorporate the effects of clinical confounding factors into the integrative analysis. Here we report a deep learning setup, named Autoencoder-based Integrative Multi-omics data Embedding (AIME), to extract data representation for omics data integrative analysis. The method can adjust for confounder variables, achieve informative data embedding, rank features in terms of their contributions, and find pairs of features from the two data types that are related to each other through the data embedding. In simulation studies, the method was highly effective in the extraction of major contributing features between data types. Using two real microRNA-gene expression datasets, one with confounder variables and one without, we show that AIME excluded the influence of confounders, and extracted biologically plausible novel information. The R package based on Keras and the TensorFlow backend is available at https://github.com/tianwei-yu/AIME.