A methylation array analysis pipeline tailored for discovering rare methylation events with interactive data visualization.
A differentially methylated region (DMR) is a genomic region that has different DNA methylation status across samples. Rare DMRs have been shown to cause multiple diseases. A famous example is Fragile X sydrome where the majority of patients display a hypermethylated DMR in the 5'UTR of FMR1 leading to reduced gene expression. The hypermethylation is driven by an underying GC-rich expansion repeat that is impossible to detect by short-read sequencing approaches. Therefore, it is very useful to "mine" these rare methylation events in order to discover novel causes of disease.
- Christy LaFlamme
- Pandurang Kolekar
- Nadhir Djekidel
- Wojciech Rosikiewicz
Methylation array analysis pipeline consists of two major components:
- Processing of data using NextFlow pipeline
- Interactive exploration using Jupyter Dash server
Current implementation of the pipeline is designed to be executed on St. Jude HPCF Research Cluster. However, the code can be run from anywhere as long as the <nextflow.config> file is adjusted and the dependencies are installed/available:
- nextflow (v21.10.5)
- Perl (v5.26.3)
- R (v4.1.3)
- BiocManager
- minfi
- IlluminaHumanMethylationEPICmanifest
- IlluminaHumanMethylationEPICanno.ilm10b4.hg19
- doParallel
- funr
- argparse
- tidyverse
- ggprism
- reshape2
- pheatmap
- rstudioapi
HPCF: After logging into interactive node, please load the following or latest available NextFlow module before running the pipeline
module load nextflow/21.10.5
- workdir - Input directory with *.idat files and single *.csv (description in below section) file with description of the samples
- runName - Name for the execution run of the pipeline
- outdir - Output directory for the pipeline to save the output files/folders
- genome - genome for QC/DMR analysis (currently only support hg19)
- platform - Methylation array platform (currently only support EPIC)
- windowSize - Window size (bp) for DMR analysis (default: 1000)
- qCutMin - Minimum quantile cut-off for DMR analysis (default: 0.25)
- qCutMax - Maximum quantile cut-off for DMR analysis (default: 0.75)
- email - Email address for notification of the pipeline status
- Sample_Name - Column containing the sample names
- Sentrix_ID - Should be first portion of unique identifier consisting of 12 numbers (e.g. "204776850065")
- Sentrix_Position - Should be row column number (e.g. R01C01)
- Sample_ID - Should be the Sentrix_ID and the Sentrix Position corresponding to the full unique identifier (e.g. "204776850065_R01C01")
- Reported_Sex - Options: Male, Female, Unknown (required if sex check for QC is desired)
- Sample_Group - Option: Case, Control (required if post-analysis filtering is desired)
Columns must be named exactly as indicated above. Also, .csv file must end on a new line (return character). And be careful! Microsoft Excel likes to convert Sentrix IDs to scientific notation.
- Sample_Type - Column containing the sample tissue source (e.g. Blood, Saliva, Fibroblast, etc.)
- Sentrix_Well - Well number corresponding to the sample plate submission
- Sample_Plate - Plate name that sample was run on (good annotation to check for batch effects between plates)
- Date - Date of sample run
- Directory-level output organization
<outdir>
└── <runName>
└── <runName>_preprocessIllumina
├── normalized_data
│ ├── betaValues
│ ├── bigWig
│ │ ├── autosomes
│ │ ├── female
│ │ └── male
│ ├── dmr_<windowSize>_<qCutMin>_<qCutMax>
│ └── qcReports
├── raw_data
└── sex_prediction
- Directory/File-level output organization
<outdir>
└── <runName>
└── <runName>_preprocessIllumina
├── normalized_data
│ ├── betaValues
│ │ ├── autosomes.beta.txt.sorted
│ │ ├── female.sexchr.beta.txt.sorted
│ │ └── male.sexchr.beta.txt.sorted
│ ├── bigWig
│ │ ├── autosomes
│ │ │ ├── <sample_1>.bed
│ │ │ ├── <sample_1>.bw
│ │ │ ├── <sample_2>.bed
│ │ │ ├── <sample_2>.bw
│ │ │ ├── <sample_n>.bed
│ │ │ └── <sample_n>.bw
│ │ ├── female
│ │ │ ├── <sample_1>.bed
│ │ │ ├── <sample_1>.bw
│ │ │ ├── <sample_2>.bed
│ │ │ ├── <sample_2>.bw
│ │ │ ├── <sample_n>.bed
│ │ │ └── <sample_n>.bw
│ │ └── male
│ │ ├── <sample_1>.bed
│ │ ├── <sample_1>.bw
│ │ ├── <sample_2>.bed
│ │ ├── <sample_2>.bw
│ │ ├── <sample_n>.bed
│ │ └── <sample_n>.bw
│ ├── dmr_<windowSize>_<qCutMin>_<qCutMax>
│ │ ├── [autosomes|sexchr].beta.txt.sorted_<windowSize>_<qCutMin>_<qCutMax>_findEpivariation.txt
│ │ ├── [autosomes|sexchr].beta.txt.sorted_<windowSize>_<qCutMin>_<qCutMax>_findEpivariation.txt.sig
│ │ └── [autosomes|sexchr].beta.txt.sorted_<windowSize>_<qCutMin>_<qCutMax>_findEpivariation.txt.sig_dmr.txt
│ └── qcReports
│ ├── beta.density.bygroup.[autosomes|sexchr].pdf
│ └── beta.density.bysample.[autosomes|sexchr].pdf
├── phenoData.all.RDS
├── raw_data
│ ├── density_beta_raw.pdf
│ ├── plotQC_raw.pdf
│ ├── qcReport.pdf
│ └── RGSet.all.RDS
└── sex_prediction
├── sexplot.pdf
├── sexPredictionsComparison.discordant.txt
└── sexPredictionsComparison.txt
Clone the repository on your cluster directory and navigate into a pipeline directory, methFlow.
The default cluster parameters in the nextflow.config file can be modified suitably. Please refer to the configuration section of NextFlow Documentation for the same.
Pipeline consists of the following three workflows.
-
QC_WF
-
Workflow to perform only Quality Checks (QC) on input raw data
-
Command format:
nextflow run map.nf -entry QC_WF \ --workdir <Path/to/WorkDir> \ --runName <RunName> \ --outdir <Path/to/OutDir> \ --email <email address>
-
-
DMR_WF
-
Workflow to perform Differentially Methylated Region (DMR) analysis on output generated by QC_WF workflow
-
Command format:
nextflow run map.nf -entry DMR_WF \ --workdir <Path/to/WorkDir> \ --runName <RunName> \ --outdir <Path/to/OutDir> \ --windowSize <num> \ --qCutMin <qCutMin> \ --qCutMax <qCutMin> \ --email <email address> -
Note
- DMR_WF workflow can be run multiple times on the same QC output with different window size and quantile cut-off parameters.
- Use the same runName and outdir parameters as for QC_WF
- Pipeline will generate separate output directories prefixing the related parameters
-
-
QUANTILE_WF
-
Workflow to perform QC abd DMR analysis on input raw data
-
Command format:
nextflow run map.nf -entry QUANTILE_WF \ --workdir <Path/to/WorkDir> \ --runName <RunName> \ --outdir <Path/to/OutDir> \ --windowSize <num> \ --qCutMin <qCutMin> \ --qCutMax <qCutMin> \ --email <email address>
-
Place the app.py file from MethVis subfolder inside the <runName>_preprocessIllumina directory, activate the environment with all dash-related packages (see installation instructions above), and run command python app.py.
- Support for case vs. control analysis
- Support for different genomes (hg38, mm9, mm10)
- Support for different methylation array platforms (450K, 935K EPICv2 etc)
- Support for different normalization methods (preprocessNoob etc)
- Support for cell type composition QC analysis (estimateCellCounts)
- Support for CNV calling using (conumee)
- Docker/Singularity containerization for cross-platform execution with required dependencies
We recommend generating the Anaconda virtual environment, to install the required packages. Assuming this is the approach taken, first create and activate the enviroment:
conda create -n dash
conda activate dash
then install all dependencies inside:
conda install jupyterLab numpy scipy matplotlib seaborn pandas matplotlib-venn ipykernel plotly dash dash-bio dash-core-components dash-bootstrap-components notebook jupyter-dash -c conda-forge -c anaconda -c plotly
$ cp ./MethVis/app.py /path/to/<runName>_preprocessIllumina/
$ cd /path/to/<runName>_preprocessIllumina/
$ conda activate dash
$ python app.py
The script will process the output files and output URL for the hosting Jupyter Dash server.
- Option to automatically run the server at the end of pipeline execution
- More annotation tracks and interactive visualization plots
Six (6) .idat files corresponding to three (3) Fragile X samples are made available as test data for the MethMiner pipeline. These samples serve as positive controls for DMR calling and sample data visualization. Three (3) samples (2 male Fragile X patients, 1 female Fragile X patient) are positive controls for a hypermethylated DMR upstream FMR1 gene on ChrX. The metadata for these samples can be found in the same folder as the ./data/FragileX_exampleData/.
DMR calling : Garg P, Jadhav B, Rodriguez OL, Patel N, Martin-Trujillo A, Jain M, Metsu S, Olsen H, Paten B, Ritz B, Kooy RF, Gecz J, Sharp AJ. A Survey of Rare Epigenetic Variation in 23,116 Human Genomes Identifies Disease-Relevant Epivariations and CGG Expansions. Am J Hum Genet. 2020;107(4):654-69. Epub 2020/09/17. doi: 10.1016/j.ajhg.2020.08.019. PubMed PMID: 32937144; PMCID: PMC7536611.
DMR annotations : Zhou W, Laird PW, Shen H. Comprehensive characterization, annotation and innovative use of Infinium DNA methylation BeadChip probes. Nucleic Acids Res. 2017;45(4):e22. Epub 2016/12/08. doi: 10.1093/nar/gkw967. PubMed PMID: 27924034; PMCID: PMC5389466.