Skip to content Skip to sidebar Skip to footer
Página inicial / Calculate Read Depth of Regions in Bisulphite Sequenced Data

Calculate Read Depth of Regions in Bisulphite Sequenced Data

Postar um comentário

Epigenetic phenomena are heritable changes in cellular phenotype that are not due to mutations in DNA sequence (1, 2). Mechanisms responsible for these phenomena include noncoding RNA species, covalent modifications of histone proteins, and methylation of DNA cytosine residues (three). In animals, Deoxyribonucleic acid methylation predominantly, although not exclusively (iv), occurs at cytosine residues that are followed by a guanine balance on their 3′ flank, referred to as cytosine-phospho-guanine (CpG) dinucleotides to distinguish them from CG interstrand base pairing (5). Approximately 4% of cytosines appear in CpG context, and lx–80% of CpG cytosines are methylated depending on the cell type and physiologic or pathologic country. Importantly, CpG residues tend to showroom a highly nonuniform distribution, clustering together in so-called CpG islands, which are defined as >200-bp regions (typically ∼1 kb) with a GC fraction greater than 0.v and an observed-to-expected CpG ratio greater than 0.6 (6). These CpG islands localize well-nigh gene promoters and other gene-regulatory elements, and tend to be hypomethylated (7) (Figures 1A–1D ). In full general, CpG methylation is associated with transcriptional repression, and the clan is causal in many, but not all, contexts. Moreover, only as the distribution of CpG dinucleotides tends toward nonuniformity, their pattern of methylation is also nonuniform, with methylated residues clustering together in patterns that can vary dramatically betwixt prison cell types, functional states, and affliction atmospheric condition.

DNA methylation is a chemically stable yet biologically dynamic mark (Figure 2A ). A family of DNA methyltransferases (DNMTs) catalyzes the conversion of cytosine to 5-methylcytosine (8). Maintenance of Deoxyribonucleic acid methylation during cell division is a regulated process involving the DNA methyltransferase Dnmt1 and its regulatory adapter protein Uhrf1, among other molecules (9, 10). De novo methylation occurs via Dnmt3a or Dnmt3b activity. Deoxyribonucleic acid demethylation tin can occur passively during DNA replication or via the catalytic activity of the ten-eleven translocase (TET) family of dioxygenase enzymes (11). Although information technology was once considered to exist a long-lasting epigenetic marking because of its thermodynamic stability, Dna methylation and demethylation are in fact quite dynamic, with rapid kinetics observed in multiple biologic systems (12–14).

Numerous developmental, physiologic, and pathologic processes exhibit specific DNA methylation patterns (15). These processes include the development of myriad cell types and tissues, the plasticity of immune cell identity and function, and malignancy. Because of the power inherent in epigenetic control mechanisms, researchers have adult sophisticated tools to investigate Dna methylation in both animal models and human subjects. My goal hither is to provide a focused overview of technologies and computational strategies to mensurate and analyze Deoxyribonucleic acid methylation, highlighting bisulfite sequencing-based methods and pipelines, and using some of my group's techniques and informatics procedures to illustrate central concepts. This review is not intended to be comprehensive, simply rather to serve as a practical guide to explore DNA methylation measurement and data analysis. Excellent reviews and guidelines on adjacent-generation sequencing take been published recently (16–18). The reader is encouraged to review these references for a thorough background on adjacent-generation sequencing, control of batch furnishings, and other problems relevant to the design and analysis of sequencing-based studies.

Methods of Measuring Deoxyribonucleic acid Methylation

Section:

Previous section Next section

Numerous technologies allow measurement of DNA methylation. Each has its own advantages and disadvantages, and these are reviewed in depth in Reference 19 and summarized in Tabular array 1. Most mutual methods involve a treatment that distinguishes unmethylated from methylated cytosines, followed by a step that leverages this identification strategy to generate a DNA methylation information set. Although most of this review volition focus on methods that use a chemical strategy to distinguish unmethylated from methylated cytosines followed by side by side-generation sequencing (bisulfite sequencing), information technology is important to hash out other common techniques, such as affinity enrichment methods. Strategies that exploit methylation-sensitive restriction endonucleases coupled with assortment hybridization (e.yard., comprehensive loftier-throughput arrays for relative methylation [Charm] [xx]) and adjacent-generation sequencing [Methyl-seq] [21]), as well equally unmarried-cell approaches (22) and single-molecule nanopore sequencing (23), may be of interest but are not reviewed farther here.

Tabular array 1. Common Techniques Used to Measure DNA Methylation
Technique Advantages Disadvantages Notes
Affinity enrichment based (eastward.chiliad., MeDIP and MBD-based methods) Low cost relative to bisulfite sequencing. Low resolution relative to bisulfite sequencing. Bias due to copy number variation, GC content, and CpG density. Higher input requirements than bisulfite conversion–based methods. Straightforward for laboratories already facile with chromatin immunoprecipitation, sequencing chemical science, and bioinformatics.
Bisulfite conversion based (e.grand., WGBS, RRBS, and Infinium) Depression input requirements (pg–ng calibration). Single-nucleotide resolution. Can provide non-CpG information. Labor and computation intensive compared with affinity enrichment techniques. Susceptible to bias from incomplete bisulfite conversion and bisulfite PCR artifacts. Current gilded standard. Requires specialized chemistry and computational platforms. Oxidative bisulfite sequencing permits identification of 5-hydroxymethylcytosine (run into text for alternatives).

Analogousness Enrichment Strategies

Common affinity enrichment strategies include methyl-Dna immunoprecipitation (MeDIP (24–26)) and methyl-CpG binding domain poly peptide (MBD [27, 28]) methods. Figure 2B illustrates the bones steps involved in these techniques. Both MeDIP- and MBD-based methods accept the advantage of being low-toll and straightforward for laboratories that are already skilled in chromatin immunoprecipitation sequencing (Flake-seq) or related techniques. The disadvantages include relatively depression resolution, susceptibility to copy number variation bias, and a bias toward observations of methylated Deoxyribonucleic acid compared with bisulfite-based protocols (19).

Bisulfite-based Methods

Although bisulfite-based methods are more labor and ciphering intensive than other approaches, many consider them to exist the gold standard for measuring Dna methylation because of their unmarried-nucleotide resolution, flexibility beyond organisms and model systems, and very low input requirements (we take successfully performed bisulfite sequencing on 10–100 pg of genomic DNA). Equally detailed in Figure 2C , treatment of genomic DNA with sodium bisulfite transforms epigenetic information into genetic information that tin can then be assessed with the use of strategies detailed below. The fundamental consequence of the bisulfite conversion reaction is rapid transformation of unmethylated cytosine residues to uracil residues—a reaction from which 5-methylcytosine residues are thermodynamically protected (29, xxx). It is critical to achieve very high cytosine-to-uracil conversion rates to satisfy the assumptions of bisulfite-based assay discussed below; our conversion rates are routinely greater than 99%, as measured by the observed frequency of unmethylated CpGs in an unmethylated λ-bacteriophage genome spiked into every sample. Bisulfite-converted Dna is fragmented considering of the harsh chemic treatment. It is as well unmarried stranded because the supermajority of not-CpG cytosines are unmethylated and thus are converted to uracils, resulting in loss of CG interstrand base pairing. Importantly, when these bisulfite-converted fragments are subjected to standard PCR amplification, thymines replace uracils (Figure 2D ), creating a DNA sequence that can be compared with a reference unconverted sequence to make up one's mind whether individual cytosines were methylated or not in the original sample. The PCR amplification tin be locus specific, in which case the amplified fragments are cloned and sequenced past standard Sanger-based methods or pyrosequencing, or subjected to targeted deep-amplicon bisulfite sequencing. Methylation-specific PCR assays can also be designed to amplify but converted (i.due east., thymine-ated) sequences, thus distinguishing methylated from unmethylated genomic regions of interest. Forth with mass spectrometry–based methods such as the EpiTYPER platform (31), these locus-specific techniques can as well be used to validate findings obtained from genome-wide methods.

Genome-scale interrogation of methylation status at single-nucleotide resolution can exist performed via array hybridization of bisulfite-converted Deoxyribonucleic acid using site-specific, bead-ligated probes that distinguish methylated and unmethylated loci based on their differential sequence after bisulfite treatment. The most recent iteration of the commonly used Illumina Infinium methylation assay uses this arroyo to measure methylation at up to 850,000 sites (32) and is popular for big-scale human studies. Comprehensive methylation profiling can be performed with whole-genome bisulfite sequencing (WGBS), which represents the current gold standard for Deoxyribonucleic acid methylation cess (33). In WGBS, strategies such as random PCR priming are used to amplify Dna without respect to whatsoever specific loci. Adapter ligation and indexing (barcoding) tin occur earlier or after bisulfite conversion, and these adapter-ligated fragments are then sequenced using next-generation technologies. Later on the data-processing steps outlined below, a computer algorithm assigns an unmethylated value to sequenced thymines occurring at positions for which the reference genome contains a cytosine. Conversely, a methylated value is assigned to sequenced cytosines occurring at positions for which the reference genome contains a cytosine. For some and so-called nondirectional protocols, including ours, all four strands that result from bisulfite treatment—the original top, original bottom, complement to the original tiptop, and complement to the original bottom—are sequenced and provide methylation information (see Figure 2nd ). Thus, in addition to detecting cytosine-to-thymine conversions, sophisticated algorithms can find unmethylated residues past recording guanine-to-adenine conversions (i.due east., the result of a cytosine-to-thymine conversion on the complementary strand), as illustrated in Figure 2D and discussed below.

WGBS provides the nearly comprehensive assessment of cytosine methylation, although knowing the methylation status of about every genomic cytosine in any context (not only CpG) is unnecessary for virtually studies. Moreover, as cytosines tend to brandish locally conserved methylation status, it is as well non typically necessary to measure the methylation status of every CpG because the methylation status of nearby cytosines tin exist inferred. Accordingly, our group and many others perform reduced representation bisulfite sequencing (RRBS), which implements an initial unsupervised enrichment step for CpG-rich regions of the genome (34–38). Our modified RRBS (mRRBS) protocol is illustrated in Figure 2E . Although the technical details vary, most RRBS procedures measure 10–twenty% of all genomic CpGs (upwards of ii–4 million CpGs in mice or humans) while sequencing only 1–ii% of the total genome because of the critical digestion and enrichment steps. This approach produces toll savings in terms of sequencing expenses and enables multiplexing of multiple indexed (barcoded) samples into a sequencing run to limit batch effects. For comparison, the NIH Roadmap Epigenomics Project'southward guidelines for WGBS (http://www.roadmapepigenomics.org/protocol) advise a 30× depth at the whole-genome calibration and a minimum of 100-bp reads (>800–i,000 million aligned reads in full), whereas nosotros target ∼l million aligned reads per mRRBS sample. Accordingly, we multiplex 4 to half-dozen samples per run using single-end 75-bp reads on an Illumina NextSeq 500 instrument with a V2 Loftier-Output Reagent Kit (∼400 million reads/sequencing run). This flexibility allows the incorporation of additional biological replicates, which increases the statistical power of bisulfite sequencing studies. Additional replicates and increased sequencing depth improve the detection rate of differentially methylated loci for a given deviation in methylation, and implementation guidelines can exist plant in Reference 39. It is important to note that standard bisulfite-based techniques cannot distinguish v-methylcytosine from other 5-carbon cytosine modifications, including v-hydroxymethylation. Techniques such as oxidative bisulfite sequencing (xl), Tet-assisted bisulfite sequencing (41), hydroxy-MeDIP (which is similar to MeDIP just uses anti-5-hydroxymethylcytosine antibodies) (42), and selective 5-hydroxymethylcytosine labeling techniques, such equally a combined glycosylation restriction analysis (43, 44), are available to measure five-hydroxymethylation but are beyond the telescopic of this review.

Building a Bisulfite Sequencing Data-Processing Pipeline

Section:

Previous section Next section

The goal of bioinformatics pipelines is to provide reproducible processing of sequencing information, generating the same output for a given raw data set, pipeline components, and input variables. Many pipelines and pipeline components for processing and analyzing DNA methylation data have been published (45–49). In this department, my objective is to illustrate the general contours of a bisulfite-based processing pipeline by reviewing the steps we employ to procedure our WGBS and mRRBS data. Our pipeline, written for command line, contains modules for demultiplexing, quality assessment, trimming, alignment to reference genomes, and finally methylation extraction (also known every bit methylation calling) (Effigy 3A ). Pipelines for processing bisulfite sequencing information are computation, memory, and storage-space intensive; use of a high-operation computing cluster or cloud-based calculating system is recommended. Example commands from our pipeline are included in the data supplement.

Demultiplexing

The standard output of Illumina sequencers consists of base call (*.bcl) files. Specially when multiple uniquely indexed samples are sequenced together, information technology is necessary to create quality-annotated sequence files (*.fastq files) for each sample. Different the other steps of our pipeline, demultiplexing bisulfite sequencing data requires no special modifications to standard packages such as Illumina's BCL2FASTQ software (https://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.html). Afterward running BCL2FASTQ, it is useful to review the demultiplexing quality statistics, including the relative proportion of each indexed library to ensure even representation of each library in the pool.

Quality Assessment

Nosotros perform a multidimensional quality assessment of *.fastq files both before and after the trimming procedure outlined below. Our pipeline uses FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) to measure multiple aspects of *.fastq file quality. It is valuable to review the total number of reads obtained per sample; once again, nosotros aim for a minimum of fifty million aligned reads per sample for mRRBS. The per-base sequence quality graph is also useful to ensure good quality (average quality score >28–thirty across read positions). The per-base of operations sequence content metric, which reports the relative frequency of each DNA base across read positions, will invariably neglect considering of the bisulfite treatment, which disproportionately increases thymines (and adenines in nondirectional libraries) in comparison with other bases. Although it is reported as a failure, the observation of a disproportionate frequency of bases across read positions is a coarse indicator of the success of bisulfite conversion. GC ratios and content measurements are too confounded past bisulfite treatment, and sequence duplication levels are often college than expected in RRBS due to the enrichment for CpG-rich portions of the genome. Post-trimming FastQC reports are useful to decide the number of reads that remain afterwards trimming, ensure removal of adapters, and verify that whatsoever variability in per-base sequence content imparted by random priming has been removed (encounter below). Instance pre- and mail-trimming FastQC reports are available in the data supplement.

Trimming

Our pipeline uses Trim Galore! (https://www.bioinformatics.babraham.ac.united kingdom of great britain and northern ireland/projects/trim_galore/), a wrapper effectually Cutadapt (https://github.com/marcelm/cutadapt/) and FastQC, which has useful features for trimming *.fastq files generated from bisulfite sequencing experiments. For example, in contrast to many other trimming packages, Trim Galore! allows the states to specify that our mRRBS libraries are generated from MspI-digested fragments. Later on adapter trimming, this option instructs the software to remove some other 2 bp from the iii′ finish to avoid an artifact introduced during grooming of MspI-digested libraries. We also specify the nondirectional nature of our libraries (i.east., they must include all four possible strands resulting from bisulfite treatment and amplification, not just the original top and original bottom). In addition to standard adapter trimming and trimming of depression-quality bases, we also trim bases from the five′ end of each read to remove the artifacts added by random priming (annotation the departure in the per-base of operations sequence content before and after trimming in the example FastQC reports found in the data supplement).

Alignment

The challenge with aligning bisulfite sequencing reads comes from the fact that every sequenced thymine could represent either a genuine genomic thymine or a bisulfite-converted cytosine. Too, on the complementary strand, every adenine could correspond either a genuine genomic adenine or the complement to a thymine that resulted from bisulfite conversion of an unmethylated cytosine. Therefore, alignment and methylation extraction of bisulfite sequencing data require not only a standard reference genome but besides an in silico bisulfite-converted genome, including both cytosine-to-thymine and guanine-to-adenine versions to account for the original and complementary strands, respectively (Figure 3B ). We use the popular Bismark packet (fifty) for multiple steps of our pipeline, including genome conversion, which must be completed once before alignment. We selected Bismark, which uses the Bowtie two alignment algorithm (51), equally our standard aligner considering of its integrated features and relative resistance to error across ranges of methylation levels compared with other packages (52).

Our pipeline executes two alignment scripts for each sample, creating aligned, sorted, and indexed *.bam files: one for alignment to the genome corresponding to the experiment (unremarkably mouse or human) and i to the ∼48-kb λ-bacteriophage genome added to every sample earlier bisulfite conversion. The result is a Bismark alignment report, which summarizes numerous important parameters, including the mapping rate, which is typically lower in bisulfite sequencing than other sequencing technologies due to the complexities of alignment as discussed higher up, and an estimate of the methylation frequency in each possible cytosine context (CpG, CHG, and CHH, where H is whatsoever noncytosine base of operations). The frequency of CpG methylation in the λ-bacteriophage genome gives an estimate of the bisulfite conversion efficiency, as the λ-bacteriophage genome is grown in methylase-negative Escherichia coli, which cannot methylate cytosines in CpG context. Thus, for instance, if the observed CpG methylation frequency in the λ-bacteriophage is 0.ii%, the cytosine-to-thymine (bisulfite) conversion in the sample is estimated to be 99.8% efficient.

Methylation Extraction

The final footstep in our processing pipeline also uses Bismark to perform methylation extraction. The principle is straightforward: assign a methylated call when a cytosine is observed at a position showing a cytosine in the reference genome, and assign an unmethylated phone call when a thymine is observed at a position showing a cytosine in the reference genome (Figure 3C ). This process is iterated across the genome, generating a number of outputs, including raw methylation call files for each cytosine context and strand (CpG, CHG, and CHH for the 2 original and ii complementary strands), *.bedgraph tracks for visualization of methylation at each position in standard genome browsers, a methylation call report with associated M-bias plots, and a methylation coverage file. One thousand-bias plots are useful to decide whether any substantial bias exists in methylation calls across reads (see case in the data supplement). The methylation coverage (*.cov) file is the most useful format for analysis, as it lists the methylation percentage in improver to the total number of methylated and unmethylated calls for each CpG positon.

Statistical Hypothesis Testing for Differential DNA Methylation

Department:

Previous section Next section

Quantification of CpG Methylation

A useful parameter known equally β represents the boilerplate methylation at unique cytosines measured in the population of cells that brand upwardly a sample (Figures 4A–4D ). If a cytosine residue is completely unmethylated in the population, and so β = 0 (or 0%); if it is completely methylated, then β = ane (or 100%). Fundamentally, in a unmarried cell on one allele, an individual cytosine is either unmethylated or methylated, prompting the question of how β tin range continuously from 0 to 1. There are at to the lowest degree three explanations. First, β is calculated by summing the methylated calls from the methylation extraction procedure divided by the total number of reads at that position. For example, if three methylated calls and one unmethylated phone call are observed at a position covered by 4 reads, then β = 0.75 (Figure 4E ). Second, incomplete bisulfite conversion volition outcome in intermediate β scores as an artifact of uneven bisulfite conversion. 3rd, at that place may exist heterogeneity in methylation due to mixtures of prison cell types or cell states within the population used every bit a sample. If a sample contains 50% cells that are methylated at a certain cytosine position and 50% cells that are unmethylated at that position, and then β will be 0.5 if all other variables are equal. Menses-cytometric enrichment for prison cell types of interest can reduce this heterogeneity, although fixation protocols can dethrone DNA and increase the heterogeneity of DNA methylation (53). It is important to notation that although information technology is the near useful parameter to describe cytosine methylation, β can demonstrate substantial heteroscedasticity (i.e., unequal variability across its range) and does not account for read depth (i.eastward., observability) at a position (54). A few approaches can mitigate the observability consequence, including filtering for positions that accept at least a minimal depth (e.g., 3×, 5×, etc.) and using algorithms that business relationship for read depth when comparing cytosines betwixt samples. Heteroscedasticity is a more complicated upshot that tin be addressed by modeling procedures to apply mathematical transforms to β. We use the bisulphite feature methylation pipeline within the SeqMonk package (https://world wide web.bioinformatics.babraham.air-conditioning.uk/projects/seqmonk/) to perform raw quantification of β scores from *.cov files generated in the final footstep of our processing pipeline. SeqMonk also serves as a genome browser and contains helpful visualization, data-handling, and statistical procedures for bisulfite sequencing information. Our statistical hypothesis-testing procedure using the Dispersion Shrinkage for Sequencing (DSS) R/Bioconductor package discussed below also generates methylation values based on a modeling approach (55–57). Although information technology is computationally intensive, nosotros selected the DSS procedure because of its ability to account for both technical (coverage) and biological (methylation level) variability in an algorithmic, unsupervised manner that does not crave the setting of an arbitrary read depth cutoff (36, 58). DSS also uses mathematical transforms to limit the heteroscedasticity inherent in raw β scores.

Comparing Methylation between Samples

One time the raw or transformed β scores are calculated, statistical hypothesis testing can exist performed at unmarried-CpG resolution to identify CpGs that are differentially methylated betwixt groups of samples—so-called differentially methylated cytosines (DMCs). The naught hypothesis for these tests is that there is no divergence in β between groups at a given position. Many methods are available for statistical hypothesis testing, and each has its own strengths and weaknesses (58). The nigh straightforward approach involves Fisher's exact examination or the chi-foursquare test; nevertheless, these tests do not take into account biological variability or variability due to sequencing depth, and demonstrate skewing of P values toward lower-than-expected values when tested against the nix condition. A different approach that accounts for read depth and biological dispersion is based on the commonly used edgeR method for RNA sequencing (RNA-seq) and demonstrates reasonable functioning in examination settings (59).

Because methylation data are inherently bimodal (i.e., nearly β scores are near 0 or 1, as explored in Figures 4A–4D ), methods that use the binomial or β-binomial distribution tend to exhibit better functioning for methylation data than statistical tests that use other distributions. We use the DSS package to generate P values and and so a standard Benjamini-Hochberg correction for multiple comparisons to generate false discovery charge per unit (FDR) q-values at well-observed CpG positions equally divers by the DSS modeling procedure. A DMC can then be divers as a CpG with an FDR q value less than a desired threshold, typically 0.05 or 0.01. A Δ value tin exist assigned to obtain a listing of DMCs that are dissimilar by a divers magnitude in β score—for example, a β difference of 0.10 or 0.25 (i.e., 10% or 25%). The DSS procedure for pairwise comparisons involves a Bayesian hierarchical model that approximates and shrinks CpG site-specific dispersions, which accounts for both biological variability and sequencing depth (55–57). DSS then solves the β-binomial model and tests the aught hypothesis of equal mean methylation between two groups using Wald tests. DSS can besides perform hypothesis testing using F tests in a general experimental design, which allows comparison of multiple groups, factors, or other variables using a β-binomial regression model. Considering of the style in which regression coefficients are calculated in DSS, the general experimental design procedure does not quantitate β scores, but it does generate a list of well-observed positions. Accordingly, we frequently apply raw β scores at these positions when comparing multiple groups, once again avoiding the demand to use an arbitrary read depth cutoff that results in loss of information.

A list of DMCs then permits the generation of a gear up of differentially methylated regions (DMRs). The definition of a DMR is not standardized, and at that place are no well-validated procedures for generating an unsupervised set of DMRs. For example, the default DSS approach defines a DMR equally a region with a minimum length of 50 bp that contains at least 3 CpGs, 50% of which see an arbitrary P value threshold. These regions are merged when they occur within 50 bp of 1 another, creating larger DMRs without an upper jump. Consistent with the capricious definition of a DMR, the DSS packet documentation states, "It is very difficult to select a natural and rigorous threshold for defining DMRs. We recommend users try different thresholds to obtain satisfactory results." Our general approach is to ascertain regions of interest based on prior annotations of promoters, enhancers, and other functional genomic elements, then interrogate these areas for DMCs. In a converse arroyo, nosotros also examination the likelihood of DMCs actualization in a selected fix of regions (eastward.grand., promoters, enhancers, etc.) above that expected past take chances using hypergeometric testing (35). Machine learning approaches to DMR identification hold hope for the hereafter of unsupervised DNA methylation analysis (threescore).

Functional Enrichment Assay and Integration with Other –Omics Data Sets

Section:

Previous section Next section

Functional Enrichment Assay

Before integration with other –omics data sets, such as those generated past ChIP-seq or RNA-seq, we perform a functional enrichment analysis on a list of DMCs or DMRs using tools such equally the Genomic Regions Enrichment of Annotations Tool (61). Originally designed for Chip-seq information, this tool uses stringent control for false positives to acquaintance cis-regulatory regions with input genomic coordinates, drawing from an all-encompassing set of annotated ontologies. As with whatever functional enrichment tool, we are cautious most interpreting its output considering of the inherently biased nature of functional enrichment due to the human-annotated databases from which these tools draw their biological associations.

Integration with Scrap-Seq Information

DNA methylation does not exist in a vacuum, and the power of DNA methylation sequencing lies in integration with data sets generated by other –omics technologies. Integration with Fleck-seq tin can be performed by examining DNA methylation at well-observed CpGs across putative enhancers identified by occupancy of histone 3 lysine 4 monomethylation (H3K4 me1) and other DNA-bound proteins. For instance, we recently conducted a report in which we deleted TET2 in a breast cancer cell line and queried the effect on histone modifications, DNA methylation, and transcription factor bounden (37). Our arroyo began with clustering of estrogen receptor-α ChIP-seq binding peaks to identify regions with differential bounden upon loss of TET2. After confirming TET2 occupancy at these sites in wild-type cells, we used mRRBS and the DSS process to determine that loss of TET2 resulted in Dna hypermethylation at these estrogen receptor-α/TET2–regulated regions. This integration analysis allowed united states of america to establish a model in which TET2 coactivates gene expression through Dna demethylation of critical enhancer elements.

Integration with RNA-Seq Information

Perhaps the nigh common integration occurs between DNA methylation and RNA-seq (transcriptional profiling) data, equally transcription represents the proximate readout of epigenetic control mechanisms, including DNA methylation. One straightforward approach is to examine the Deoxyribonucleic acid methylation status of the promoters of differentially expressed genes. We took this arroyo in a contempo study examining the upshot of loss of mitochondrial complex III on regulatory T cell–suppressive function (38). Using a Metagene analysis and display, we determined that loss of complex III resulted in hypermethylation of the promoters of downregulated genes, particularly those with an associated CpG island. These data supported our model in which loss of complex III results in an increase in the metabolite 50-(S)-2-hydroxyglutarate, which inhibits demethylase enzymes, including the TET family of DNA demethylases (62).

The above arroyo works well with pairwise comparisons, in which relative hyper- or hypomethylation can be easily defined between 2 groups. A claiming arises when multiple groups are examined, as was the case in our study of differential Dna methylation and transcription within sorted lung CD4+ T cells during neonatal pneumonia in mice (35). In that written report, we examined four groups in a crossed experimental pattern: neonatal and juvenile mice exposed to either PBS (control) or E. coli bacteria (pneumonia). For the analysis we created a semisupervised Deoxyribonucleic acid methylation difference-filtering algorithm, which is explored in Figures 5A–5D . Conceptually, the algorithm begins past determining the genes that are one) differentially expressed in the RNA-seq data set (by an ANOVA-similar test in edgeR [63]) and 2) differentially methylated in the mRRBS data set (liberally defined as genes with at least 1 DSS general experimental blueprint-defined DMC within 2 kb of their cistron bodies, inclusive). The overlap between these two cistron lists is evaluated using a hypergeometric exam and visualized using a simple Venn diagram. If the overlap is greater than that expected by chance, the list of overlapping genes is so subjected to k-means clustering using standard procedures (16). Based on the assumption that Deoxyribonucleic acid methylation in promoters is a repressive mark, the algorithm and then selects (filters for) CpGs within gene promoters that are hypermethylated within the lower-expressed groups (and therefore hypomethylated in the college-expressed groups) past an arbitrary divergence in the β score, commonly 0.i or 0.25 (i.e., 10% or 25%). This footstep is repeated for each k-means cluster based on the observed pattern of expression particular to that cluster. The result is a subset of genes passing the methylation filter whose promoters display a methylation design that is anticorrelated with gene expression, befitting to the biologic assumption of methylation as a repressive mark. This terminal listing of candidate genes has a loftier statistical probability of being regulated by Deoxyribonucleic acid methylation. In the neonatal pneumonia report, we used this procedure to make up one's mind a list of genes that class a core regulatory signature inside lung CD4+ T cells along both developmental (time) and response-to-inflammation (pneumonia) axes. In summary, although only a few standardized, validated approaches are available to perform unsupervised integration of Deoxyribonucleic acid methylation and other –omics information sets (64), creative strategies can exist used to combine these data sets in biologically informative ways.

Conclusions and General Recommendations

Department:

Previous section Next section

Dna methylation is a key, dynamic epigenetic mark that is involved in myriad developmental, homeostatic, and pathologic processes. A detailed mechanistic understanding of the biology of Deoxyribonucleic acid methylation every bit a biomarker or causal substrate requires methods to measure and analyze DNA methylation using depression-bias and high-resolution techniques. Although many approaches can be used to reach these goals, in this review I take highlighted bisulfite sequencing as the current golden standard, and outlined a biochemical and analytical strategy to measure and clarify Deoxyribonucleic acid methylation in a comprehensive, unmarried-nucleotide-resolution, unsupervised manner. These techniques, particularly the computational methods, may seem daunting for junior and senior investigators alike. Nevertheless, I would encourage interested readers to attempt these techniques in their own laboratory, outset at a minor scale. Numerous online tutorials and resource tin can be accessed to outset exploring DNA methylation data. The SeqMonk platform discussed to a higher place is a convenient graphical user interface accompanied by a helpful tutorial that can provide an easy entrance into the world of Deoxyribonucleic acid methylation data analysis. Going forward as technologies and computational resources evolve, adherence to sound scientific, mathematical, and computational principles will be important to facilitate discoveries involving DNA methylation in the context of other epigenetic phenomena and the numerous regulatory levels linking genome, environs, and cellular function.

The author thanks Shang-Yang (Sam) Chen and Kishore Anekalla for their work in developing the initial lawmaking sets for the Deoxyribonucleic acid methylation processing and analysis pipelines, every bit well as the members of his laboratory and research group for their helpful comments on the manuscript. This work was supported in part past the computational resources and staff contributions provided by the Genomics Compute Cluster, which is jointly supported past the Feinberg School of Medicine, the Middle for Genetic Medicine, and Feinberg'southward Department of Biochemistry and Molecular Genetics, the Role of the Provost, the Office for Research, and Northwestern Information technology. The Genomics Compute Cluster is part of Quest, Northwestern University's high-performance calculating facility, with the purpose to advance research in genomics.

References

Department:

Previous section Next section

1. Waddington CH . The epigenotype: 1942. Int J Epidemiol 2012;41:10–13.

Crossref, Medline, Google Scholar
2. Berger SL , Kouzarides T , Shiekhattar R , Shilatifard A . An operational definition of epigenetics. Genes Dev 2009;23:781–783.

Crossref, Medline, Google Scholar
3. Goldberg Advertizing , Allis CD , Bernstein E . Epigenetics: a landscape takes shape. Cell 2007;128:635–638.

Crossref, Medline, Google Scholar
4. Schultz MD , He Y , Whitaker JW , Hariharan M , Mukamel EA , Leung D , et al. Human body epigenome maps reveal noncanonical Dna methylation variation. Nature 2015;523:212–216.

Crossref, Medline, Google Scholar
5. Bird A . Deoxyribonucleic acid methylation patterns and epigenetic memory. Genes Dev 2002;16:half dozen–21.

Crossref, Medline, Google Scholar
6. Gardiner-Garden Thousand , Frommer Thousand . CpG islands in vertebrate genomes. J Mol Biol 1987;196:261–282.

Crossref, Medline, Google Scholar
7. Saxonov S , Berg P , Brutlag DL . A genome-broad analysis of CpG dinucleotides in the human genome distinguishes two distinct classes of promoters. Proc Natl Acad Sci USA 2006;103:1412–1417.

Crossref, Medline, Google Scholar
8. Lyko F . The DNA methyltransferase family: a versatile toolkit for epigenetic regulation. Nat Rev Genet 2018;19:81–92.

Crossref, Medline, Google Scholar
9. Bostick One thousand , Kim JK , Estève PO , Clark A , Pradhan S , Jacobsen SE . UHRF1 plays a function in maintaining DNA methylation in mammalian cells. Science 2007;317:1760–1764.

Crossref, Medline, Google Scholar
10. Sharif J , Muto M , Takebayashi S , Suetake I , Iwamatsu A , Endo TA , et al. The SRA protein Np95 mediates epigenetic inheritance by recruiting Dnmt1 to methylated Dna. Nature 2007;450:908–912.

Crossref, Medline, Google Scholar
11. Wu 10 , Zhang Y . TET-mediated active DNA demethylation: machinery, function and beyond. Nat Rev Genet 2017;18:517–534.

Crossref, Medline, Google Scholar
12. Pacis A , Tailleux Fifty , Morin AM , Lambourne J , MacIsaac JL , Yotova 5 , et al. Bacterial infection remodels the Deoxyribonucleic acid methylation landscape of human dendritic cells. Genome Res 2015;25:1801–1811.

Crossref, Medline, Google Scholar
13. Wallner S , Schröder C , Leitão E , Berulava T , Haak C , Beißer D , et al. Epigenetic dynamics of monocyte-to-macrophage differentiation. Epigenetics Chromatin 2016;9:33.

Crossref, Medline, Google Scholar
fourteen. Vocaliser BD , Mock JR , Aggarwal NR , Garibaldi BT , Sidhaye VK , Florez MA , et al. Regulatory T cell DNA methyltransferase inhibition accelerates resolution of lung inflammation. Am J Respir Jail cell Mol Biol 2015;52:641–652.

Abstract, Medline, Google Scholar
15. Morales-Nebreda L , McLafferty FS , Vocalizer BD . Dna methylation as a transcriptional regulator of the immune organisation. Transl Res 2019;204:1–18.

Crossref, Medline, Google Scholar
16. Koch CM , Chiu SF , Akbarpour M , Bharat A , Ridge KM , Bartom ET , et al. A beginner's guide to analysis of RNA sequencing data. Am J Respir Cell Mol Biol 2018;59:145–157.

Abstract, Medline, Google Scholar
17. Winter DR . Thinking Large rheumatology: how to make functional genomics data piece of work for you. Arthritis Res Ther 2018;20:29.

Crossref, Medline, Google Scholar
eighteen. Hersh CP , Adcock IM , Celedón JC , Cho MH , Christiani DC , Himes Exist , et al. High-throughput sequencing in respiratory, critical care, and sleep medicine research: an official American Thoracic Lodge workshop study. Ann Am Thorac Soc 2019;16:1–16.

Abstract, Medline, Google Scholar
19. Laird PW . Principles and challenges of genomewide Deoxyribonucleic acid methylation analysis. Nat Rev Genet 2010;11:191–203.

Crossref, Medline, Google Scholar
20. Irizarry RA , Ladd-Acosta C , Carvalho B , Wu H , Brandenburg SA , Jeddeloh JA , et al. Comprehensive high-throughput arrays for relative methylation (CHARM). Genome Res 2008;xviii:780–790.

Crossref, Medline, Google Scholar
21. Brunner AL , Johnson DS , Kim SW , Valouev A , Reddy TE , Neff NF , et al. Singled-out Dna methylation patterns characterize differentiated human being embryonic stem cells and developing human fetal liver. Genome Res 2009;19:1044–1056.

Crossref, Medline, Google Scholar
22. Kelsey Yard , Stegle O , Reik W . Single-cell epigenomics: recording the past and predicting the time to come. Science 2017;358:69–75.

Crossref, Medline, Google Scholar
23. Clarke J , Wu HC , Jayasinghe L , Patel A , Reid South , Bayley H . Continuous base identification for single-molecule nanopore DNA sequencing. Nat Nanotechnol 2009;4:265–270.

Crossref, Medline, Google Scholar
24. Weber One thousand , Hellmann I , Stadler MB , Ramos Fifty , Pääbo South , Rebhan M , et al. Distribution, silencing potential and evolutionary impact of promoter DNA methylation in the man genome. Nat Genet 2007;39:457–466.

Crossref, Medline, Google Scholar
25. Weber M , Davies JJ , Wittig D , Oakeley EJ , Haase One thousand , Lam WL , et al. Chromosome-wide and promoter-specific analyses place sites of differential DNA methylation in normal and transformed human cells. Nat Genet 2005;37:853–862.

Crossref, Medline, Google Scholar
26. Hsu HK , Weng YI , Hsu PY , Huang Th , Huang YW . Detection of Dna methylation by MeDIP and MBDCap assays: an overview of techniques. Methods Mol Biol 2014;1105:61–lxx.

Crossref, Medline, Google Scholar
27. Jørgensen HF , Adie K , Chaubert P , Bird AP . Engineering a high-affinity methyl-CpG-binding poly peptide. Nucleic Acids Res 2006;34:e96.

Crossref, Medline, Google Scholar
28. Aberg KA , Chan RF , Xie L , Shabalin AA , van den Oord EJCG . Methyl-CpG-binding domain sequencing: MBD-seq. Methods Mol Biol 2018;1708:171–189.

Crossref, Medline, Google Scholar
29. Hayatsu H . Discovery of bisulfite-mediated cytosine conversion to uracil, the central reaction for DNA methylation analysis—a personal account. Proc Jpn Acad, Ser B, Phys Biol Sci 2008;84:321–330.

Crossref, Medline, Google Scholar
30. Wang RY , Gehrke CW , Ehrlich M . Comparison of bisulfite modification of 5-methyldeoxycytidine and deoxycytidine residues. Nucleic Acids Res 1980;8:4777–4790.

Crossref, Medline, Google Scholar
31. Ehrich Yard , Nelson MR , Stanssens P , Zabeau M , Liloglou T , Xinarianos G , et al. Quantitative loftier-throughput analysis of Deoxyribonucleic acid methylation patterns by base of operations-specific cleavage and mass spectrometry. Proc Natl Acad Sci Us 2005;102:15785–15790.

Crossref, Medline, Google Scholar
32. Moran S , Arribas C , Esteller M . Validation of a Dna methylation microarray for 850,000 CpG sites of the human genome enriched in enhancer sequences. Epigenomics 2016;8:389–399.

Crossref, Medline, Google Scholar
33. Lister R , Pelizzola M , Dowen RH , Hawkins RD , Hon One thousand , Tonti-Filippini J , et al. Man Deoxyribonucleic acid methylomes at base of operations resolution prove widespread epigenomic differences. Nature 2009;462:315–322.

Crossref, Medline, Google Scholar
34. Meissner A , Gnirke A , Bell GW , Ramsahoye B , Lander ES , Jaenisch R . Reduced representation bisulfite sequencing for comparative high-resolution Dna methylation analysis. Nucleic Acids Res 2005;33:5868–5877.

Crossref, Medline, Google Scholar
35. McGrath-Morrow SA , Ndeh R , Helmin KA , Chen SY , Anekalla KR , Abdala-Valencia H , et al. DNA methylation regulates the neonatal CD4+ T-prison cell response to pneumonia in mice. J Biol Chem 2018;293:11772–11783.

Crossref, Medline, Google Scholar
36. Walter JM , Helmin KA , Abdala-Valencia H , Wunderink RG , Singer BD . Multidimensional cess of alveolar T cells in critically ill patients. JCI Insight 2018;3:e123287.

Crossref, Medline, Google Scholar
37. Wang Fifty , Ozark PA , Smith ER , Zhao Z , Marshall SA , Rendleman EJ , et al. TET2 coactivates cistron expression through demethylation of enhancers. Sci Adv 2018;4:eaau6986.

Crossref, Medline, Google Scholar
38. Weinberg SE , Singer BD , Steinert EM , Martinez CA , Mehta MM , Martínez-Reyes I , et al. Mitochondrial complex 3 is essential for suppressive function of regulatory T cells. Nature 2019;565:495–499.

Crossref, Medline, Google Scholar
39. Ziller MJ , Hansen KD , Meissner A , Aryee MJ . Coverage recommendations for methylation analysis past whole-genome bisulfite sequencing. Nat Methods 2015;12:230–232, 1 p following 232.

Crossref, Medline, Google Scholar
40. Booth MJ , Branco MR , Ficz G , Oxley D , Krueger F , Reik W , et al. Quantitative sequencing of 5-methylcytosine and 5-hydroxymethylcytosine at single-base resolution. Scientific discipline 2012;336:934–937.

Crossref, Medline, Google Scholar
41. Yu Thou , Hon GC , Szulwach KE , Song CX , Jin P , Ren B , et al. Tet-assisted bisulfite sequencing of 5-hydroxymethylcytosine. Nat Protoc 2012;7:2159–2170.

Crossref, Medline, Google Scholar
42. Nestor CE , Meehan RR . Hydroxymethylated Dna immunoprecipitation (hmeDIP). Methods Mol Biol 2014;1094:259–267.

Crossref, Medline, Google Scholar
43. Song CX , Szulwach KE , Fu Y , Dai Q , Yi C , Li X , et al. Selective chemical labeling reveals the genome-wide distribution of 5-hydroxymethylcytosine. Nat Biotechnol 2011;29:68–72.

Crossref, Medline, Google Scholar
44. Song CX , Yu Thou , Dai Q , He C . Detection of 5-hydroxymethylcytosine in a combined glycosylation brake assay (CGRA) using restriction enzyme Taq(α)I. Bioorg Med Chem Lett 2011;21:5075–5077.

Crossref, Medline, Google Scholar
45. Adusumalli Due south , Mohd Omar MF , Soong R , Benoukraf T . Methodological aspects of whole-genome bisulfite sequencing analysis. Brief Bioinform 2015;16:369–379.

Crossref, Medline, Google Scholar
46. Krueger F , Kreck B , Franke A , Andrews SR . Deoxyribonucleic acid methylome analysis using brusque bisulfite sequencing information. Nat Methods 2012;nine:145–151.

Crossref, Medline, Google Scholar
47. Kangeyan D , Dunford A , Iyer S , Stewart C , Hanna M , Getz G , et al. A (fire)cloud-based DNA methylation data preprocessing and quality control platform. BMC Bioinformatics 2019;xx:160.

Crossref, Medline, Google Scholar
48. Huang KYY , Huang YJ , Chen PY . BS-Seeker3: ultrafast pipeline for bisulfite sequencing. BMC Bioinformatics 2018;xix:111.

Crossref, Medline, Google Scholar
49. Graña O , López-Fernández H , Fdez-Riverola F , González Pisano D , Glez-Peña D . Bicycle: a bioinformatics pipeline to analyze bisulfite sequencing data. Bioinformatics 2018;34:1414–1415.

Crossref, Medline, Google Scholar
fifty. Krueger F , Andrews SR . Bismark: a flexible aligner and methylation caller for bisulfite-seq applications. Bioinformatics 2011;27:1571–1572.

Crossref, Medline, Google Scholar
51. Langmead B , Salzberg SL . Fast gapped-read alignment with Bowtie 2. Nat Methods 2012;9:357–359.

Crossref, Medline, Google Scholar
52. Lee JH , Park SJ , Kenta N . An integrative approach for efficient analysis of whole genome bisulfite sequencing data. BMC Genomics 2015;16(Suppl 12):S14.

Crossref, Medline, Google Scholar
53. Singer BD , Mock JR , D'Alessio FR , Aggarwal NR , Mandke P , Johnston L , et al. Flow-cytometric method for simultaneous assay of mouse lung epithelial, endothelial, and hematopoietic lineage cells. Am J Physiol Lung Cell Mol Physiol 2016;310:L796–L801.

Crossref, Medline, Google Scholar
54. Du P , Zhang X , Huang CC , Jafari Due north , Kibbe WA , Hou 50 , et al. Comparison of Beta-value and M-value methods for quantifying methylation levels by microarray assay. BMC Bioinformatics 2010;11:587.

Crossref, Medline, Google Scholar
55. Feng H , Conneely KN , Wu H . A Bayesian hierarchical model to discover differentially methylated loci from unmarried nucleotide resolution sequencing information. Nucleic Acids Res 2014;42:e69.

Crossref, Medline, Google Scholar
56. Wu H , Xu T , Feng H , Chen L , Li B , Yao B , et al. Detection of differentially methylated regions from whole-genome bisulfite sequencing data without replicates. Nucleic Acids Res 2015;43:e141.

Medline, Google Scholar
57. Park Y , Wu H . Differential methylation analysis for BS-seq information nether general experimental design. Bioinformatics 2016;32:1446–1453.

Crossref, Medline, Google Scholar
58. Zhang Y , Baheti S , Sun Z . Statistical method evaluation for differentially methylated CpGs in base resolution next-generation DNA sequencing data. Cursory Bioinform 2018;19:374–386.

Medline, Google Scholar
59. Chen Y , Pal B , Visvader JE , Smyth GK . Differential methylation analysis of reduced representation bisulfite sequencing experiments using edgeR. F1000 Res 2017;6:2055.

Crossref, Medline, Google Scholar
60. Srivastava A , Karpievitch YV , Eichten SR , Borevitz JO , Lister R . HOME: a histogram based machine learning approach for effective identification of differentially methylated regions. BMC Bioinformatics 2019;20:253.

Crossref, Medline, Google Scholar
61. McLean CY , Bristor D , Hiller M , Clarke SL , Schaar BT , Lowe CB , et al. Peachy improves functional interpretation of cis-regulatory regions. Nat Biotechnol 2010;28:495–501.

Crossref, Medline, Google Scholar
62. Singer BD , Chandel NS . Immunometabolism of pro-repair cells. J Clin Invest 2019;130:124613.

Google Scholar
63. Robinson MD , McCarthy DJ , Smyth GK . edgeR: a Bioconductor parcel for differential expression analysis of digital gene expression information. Bioinformatics 2010;26:139–140.

Crossref, Medline, Google Scholar
64. Bersanelli Yard , Mosca East , Remondini D , Giampieri E , Sala C , Castellani G , et al. Methods for the integration of multi-omics data: mathematical aspects. BMC Bioinformatics 2016;17(Suppl 2):15.

Crossref, Medline, Google Scholar

Supported past National Institutes of Health grants K08HL128867 and U19AI135964, and a Parker B. Francis Research Opportunity Award.

B.D.S. has a pending patent application—U.S. Patent App. 15/542,380, "Compositions and Methods to Accelerate Resolution of Astute Lung Inflammation."

This article has a data supplement, which is attainable from this result's table of contents at www.atsjournals.org.

Originally Published in Printing as DOI: 10.1165/rcmb.2019-0150TR on July two, 2019

Author disclosures are available with the text of this article at www.atsjournals.org.

powersafricastred.blogspot.com

Source: https://www.atsjournals.org/doi/10.1165/rcmb.2019-0150TR

Postar um comentário for "Calculate Read Depth of Regions in Bisulphite Sequenced Data"