Identifying differentially methylated genes using mixed effect and generalized least square models

HMPL: A Pipeline for Identifying Hemimethylation Patterns by Comparing Two Samples

DNA methylation (the addition of a methyl group to a cytosine) is an important epigenetic event in mammalian cells because it plays a key role in regulating gene expression. Most previous methylation studies assume that DNA methylation occurs on both positive and negative strands. However, a few studies have reported that in some genes, methylation occurs only on one strand (ie, hemimethylation) and has clustering patterns. These studies report that hemimethylation occurs on individual genes. It is unclear whether hemimethylation occurs genome-wide and whether there are hemimethylation differences between cancerous and noncancerous cells. To address these questions, we have developed the first-ever pipeline, named hemimethylation pipeline (HMPL), to identify hemimethylation patterns. Utilizing the available software and the newly developed Perl and R scripts, HMPL can identify hemimethylation patterns for a single sample and can also compare two different samples.

A novel method for identification and quantification of consistently differentially methylated regions

Advances in biotechnology have resulted in large-scale studies of DNA methylation. A differentially methylated region (DMR) is a genomic region with multiple adjacent CpG sites that exhibit different methylation statuses among multiple samples. Many so-called “supervised” methods have been established to identify DMRs between two or more comparison groups. Methods for the identification of DMRs without reference to phenotypic information are, however, less well studied. An alternative “unsupervised” approach was proposed, in which DMRs in studied samples were identified with consideration of nature dependence structure of methylation measurements between neighboring probes from tiling arrays. Through simulation study, we investigated effects of dependencies between neighboring probes on determining DMRs where a lot of spurious signals would be produced if the methylation data were analyzed independently of the probe. In contrast, our newly proposed method could successfully correct for this effect with a well-controlled false positive rate and a comparable sensitivity. By applying to two real datasets, we demonstrated that our method could provide a global picture of methylation variation in studied samples. R source codes to implement the proposed method were freely available at http://www.csjfann.ibms.sinica.edu.tw/eag/programlist/ICDMR/ICDMR.html.

A beta-mixture model for dimensionality reduction, sample classification and analysis

Background

Patterns of genome-wide methylation vary between tissue types. For example, cancer tissue shows markedly different patterns from those of normal tissue. In this paper we propose a beta-mixture model to describe genome-wide methylation patterns based on probe data from methylation microarrays. The model takes dependencies between neighbour probe pairs into account and assumes three broad categories of methylation, low, medium and high. The model is described by 37 parameters, which reduces the dimensionality of a typical methylation microarray significantly. We used methylation microarray data from 42 colon cancer samples to assess the model.

Results

Based on data from colon cancer samples we show that our model captures genome-wide characteristics of methylation patterns. We estimate the parameters of the model and show that they vary between different tissue types. Further, for each methylation probe the posterior probability of a methylation state (low, medium or high) is calculated and the probability that the state is correctly predicted is assessed. We demonstrate that the model can be applied to classify cancer tissue types accurately and that the model provides accessible and easily interpretable data summaries.

Conclusions

We have developed a beta-mixture model for methylation microarray data. The model substantially reduces the dimensionality of the data. It can be used for further analysis, such as sample classification or to detect changes in methylation status between different samples and tissues.

Preprocessing differential methylation hybridization microarray data

Background

DNA methylation plays a very important role in the silencing of tumor suppressor genes in various tumor types. In order to gain a genome-wide understanding of how changes in methylation affect tumor growth, the differential methylation hybridization (DMH) protocol has been developed and large amounts of DMH microarray data have been generated. However, it is still unclear how to preprocess this type of microarray data and how different background correction and normalization methods used for two-color gene expression arrays perform for the methylation microarray data. In this paper, we demonstrate our discovery of a set of internal control probes that have log ratios (M) theoretically equal to zero according to this DMH protocol. With the aid of this set of control probes, we propose two LOESS (or LOWESS, locally weighted scatter-plot smoothing) normalization methods that are novel and unique for DMH microarray data. Combining with other normalization methods (global LOESS and no normalization), we compare four normalization methods. In addition, we compare five different background correction methods.

Results

We study 20 different preprocessing methods, which are the combination of five background correction methods and four normalization methods. In order to compare these 20 methods, we evaluate their performance of identifying known methylated and un-methylated housekeeping genes based on two statistics. Comparison details are illustrated using breast cancer cell line and ovarian cancer patient methylation microarray data. Our comparison results show that different background correction methods perform similarly; however, four normalization methods perform very differently. In particular, all three different LOESS normalization methods perform better than the one without any normalization.

Conclusions

It is necessary to do within-array normalization, and the two LOESS normalization methods based on specific DMH internal control probes produce more stable and relatively better results than the global LOESS normalization method.

Statistical approaches for the analysis of DNA methylation microarray data

Following the rapid development and adoption in DNA methylation microarray assays, we are now experiencing a growth in the number of statistical tools to analyze the resulting large-scale data sets. As is the case for other microarray applications, biases caused by technical issues are of concern. Some of these issues are old (e.g., two-color dye bias and probe- and array-specific effects), while others are new (e.g., fragment length bias and bisulfite conversion efficiency). Here, I highlight characteristics of DNA methylation that suggest standard statistical tools developed for other data types may not be directly suitable. I then describe the microarray technologies most commonly in use, along with the methods used for preprocessing and obtaining a summary measure. I finish with a section describing downstream analyses of the data, focusing on methods that model percentage DNA methylation as the outcome, and methods for integrating DNA methylation with gene expression or genotype data.

Identifying hypermethylated CpG islands using a quantile regression model

BACKGROUND

DNA methylation has been shown to play an important role in the silencing of tumor suppressor genes in various tumor types. In order to have a system-wide understanding of the methylation changes that occur in tumors, we have developed a differential methylation hybridization (DMH) protocol that can simultaneously assay the methylation status of all known CpG islands (CGIs) using microarray technologies. A large percentage of signals obtained from microarrays can be attributed to various measurable and unmeasurable confounding factors unrelated to the biological question at hand. In order to correct the bias due to noise, we first implemented a quantile regression model, with a quantile level equal to 75%, to identify hypermethylated CGIs in an earlier work. As a proof of concept, we applied this model to methylation microarray data generated from breast cancer cell lines. However, we were unsure whether 75% was the best quantile level for identifying hypermethylated CGIs. In this paper, we attempt to determine which quantile level should be used to identify hypermethylated CGIs and their associated genes.

RESULTS

We introduce three statistical measurements to compare the performance of the proposed quantile regression model at different quantile levels (95%, 90%, 85%, 80%, 75%, 70%, 65%, 60%), using known methylated genes and unmethylated housekeeping genes reported in breast cancer cell lines and ovarian cancer patients. Our results show that the quantile levels ranging from 80% to 90% are better at identifying known methylated and unmethylated genes.

CONCLUSIONS

In this paper, we propose to use a quantile regression model to identify hypermethylated CGIs by incorporating probe effects to account for noise due to unmeasurable factors. Our model can efficiently identify hypermethylated CGIs in both breast and ovarian cancer data.