Input data

A Multi Assay Experiment object (Ramos and others 2017) is the input for all main functions of ELMER and can be generated by createMAE function.

To perform ELMER analyses, the Multi Assay Experiment needs:

  • a DNA methylation matrix or SummarizedExperiment object from HM450K or EPIC platform for multiple samples;
  • a gene expression matrix or SummarizedExperiment object for the same samples;
  • a matrix mapping DNA methylation samples to gene expression samples
  • a matrix with samples metadata (i.e. clinical data, molecular subtype information).

If TCGA data are used, the the last two matrices will be automatically generated. Based on the genome of reference selected, metadata for the DNA methylation probes, such as genomic coordinates, will be added from Wanding Zhou annotation (Zhou and others 2016); and metadata for gene annotation will be added from ensemble database (Yates and others 2015) using biomaRt (Durinck and others 2009).

DNA methylation data

DNA methylation data feeding to ELMER should be a matrix of DNA methylation beta (\(\beta\)) value for samples (column) and probes (row) processed from row HM450K array data. If TCGA data is used, processed data from GDC website will be downloaded and automatically transformed to the matrix by ELMER. The processed TCGA DNA methylation data were calculated as \(\frac{M}{(M+U)}\), where M represents the methylated allele intensity and U the unmethylated allele intensity. Beta values range from 0 to 1, reflecting the fraction of methylated alleles at each CpG in the each tumor; beta values close to 0 indicates low levels of DNA methylation and beta values close to 1 indicates high levels of DNA methylation.

If user have raw HM450K data, these data can be processed by Methylumi or minfi generating DNA methylation beta (\(\beta\)) value for each CpG site and multiple samples. The getBeta function in minfi can be used to generate a matrix of DNA methylation beta (\(\beta\)) value to feed in ELMER. And we recommend to save this matrix as meth.rda since createMAE can read in files by specifying their path which will help to reduce memory usage.

Gene expression data

Gene expresion data feeding to ELMER should be a matrix of gene expression values for samples (column) and genes (row). Gene expression value can be generated from different platforms: array or RNA-seq. The row data should be processed by other software to get gene or transcript level gene expression calls such as mapping by tophat, calling expression value by cufflink, RSEM or GenomeStudio for expression array. It is recommended to normalize expression data making gene expression comparable across samples such as quantile normalization. User can refer TCGA RNA-seq analysis pipeline to do generate comparable gene expression data. Then transform the gene expression values from each sample to the matrix for feeding into ELMER. If users want to use TCGA data, ELMER has functions to download the RNA-Seq Quantification data (HTSeq-FPKM-UQ) from GDC website and transform the data to the matrix for feeding into ELMER. It is recommended to save this matrix as RNA.rda since createMAE can read in files by specifying the path of files which will help to reduce memory usage.

Sample information

Sample information should be stored as a data.frame object containing sample ID, group labels (control and experiment). Sample ID and groups labels are required. Other information for each sample can be added to this data.frame object. When TCGA data were used, samples information will be automatically generated by createMAE function by specifying option TCGA=TRUE. A columns name TN will create the groups Tumor and Normal using the following samples to each group:

Tumor samples are:

  • Primary solid Tumor
  • Recurrent Solid Tumor
  • Primary Blood Derived Cancer - Peripheral Blood
  • Recurrent Blood Derived Cancer - Bone Marrow
  • Additional - New Primary
  • Metastatic
  • Additional Metastatic
  • Human Tumor Original Cells
  • Primary Blood Derived Cancer - Bone Marrow

Normal samples:

  • Blood Derived Normal
  • Solid Tissue Normal
  • Buccal Cell Normal
  • EBV Immortalized Normal
  • Bone Marrow Normal
## A MultiAssayExperiment object of 2 listed
##  experiments with user-defined names and respective classes. 
##  Containing an ExperimentList class object of length 2: 
##  [1] DNA methylation: RangedSummarizedExperiment with 1666 rows and 234 columns 
##  [2] Gene expression: RangedSummarizedExperiment with 3808 rows and 234 columns 
## Features: 
##  experiments() - obtain the ExperimentList instance 
##  colData() - the primary/phenotype DataFrame 
##  sampleMap() - the sample availability DataFrame 
##  `$`, `[`, `[[` - extract colData columns, subset, or experiment 
##  *Format() - convert into a long or wide DataFrame 
##  assays() - convert ExperimentList to a SimpleList of matrices
## 
## FALSE  TRUE 
##    34   234
## A MultiAssayExperiment object of 2 listed
##  experiments with user-defined names and respective classes. 
##  Containing an ExperimentList class object of length 2: 
##  [1] DNA methylation: RangedSummarizedExperiment with 1666 rows and 234 columns 
##  [2] Gene expression: RangedSummarizedExperiment with 3808 rows and 234 columns 
## Features: 
##  experiments() - obtain the ExperimentList instance 
##  colData() - the primary/phenotype DataFrame 
##  sampleMap() - the sample availability DataFrame 
##  `$`, `[`, `[[` - extract colData columns, subset, or experiment 
##  *Format() - convert into a long or wide DataFrame 
##  assays() - convert ExperimentList to a SimpleList of matrices
## A MultiAssayExperiment object of 2 listed
##  experiments with user-defined names and respective classes. 
##  Containing an ExperimentList class object of length 2: 
##  [1] DNA methylation: RangedSummarizedExperiment with 1666 rows and 234 columns 
##  [2] Gene expression: RangedSummarizedExperiment with 3738 rows and 234 columns 
## Features: 
##  experiments() - obtain the ExperimentList instance 
##  colData() - the primary/phenotype DataFrame 
##  sampleMap() - the sample availability DataFrame 
##  `$`, `[`, `[[` - extract colData columns, subset, or experiment 
##  *Format() - convert into a long or wide DataFrame 
##  assays() - convert ExperimentList to a SimpleList of matrices

Probe information

Probe information is stored as a GRanges object containing the coordinates of each probe on the DNA methylation array and names of each probe. The default probe information is fetching from Wanding Zhou annotation (Zhou and others 2016)

## GRanges object with 3 ranges and 8 metadata columns:
##              seqnames        ranges strand | address_A address_B
##                 <Rle>     <IRanges>  <Rle> | <integer> <integer>
##   cg23733394     chr1 839752-839753      + |  35610394      <NA>
##   cg23625715     chr1 976172-976173      + |  65764492  71782443
##   cg26222311     chr1 976227-976228      + |  61750445  38637435
##                  channel  designType    nextBase nextBaseRef   probeType
##              <character> <character> <character> <character> <character>
##   cg23733394        Both          II         G/A           C          cg
##   cg23625715         Grn           I           C           G          cg
##   cg26222311         Grn           I           C           G          cg
##              orientation
##              <character>
##   cg23733394          up
##   cg23625715          up
##   cg26222311          up
##   -------
##   seqinfo: 26 sequences from an unspecified genome; no seqlengths

Gene information

Gene information is stored as a GRanges object containing coordinates of each gene, gene id, gene symbol and gene isoform id. The default gene information is the ensembl gene annotation fetched from biomaRt by ELMER function.

## GRanges object with 3808 ranges and 2 metadata columns:
##                   seqnames              ranges strand | external_gene_name
##                      <Rle>           <IRanges>  <Rle> |        <character>
##   ENSG00000188984     chr1   12776118-12788726      + |            AADACL3
##   ENSG00000204518     chr1   12704566-12727097      + |            AADACL4
##   ENSG00000108270    chr17   35306175-35414171      + |               AATF
##   ENSG00000198691     chr1   94458393-94586688      - |              ABCA4
##   ENSG00000135776     chr1 229652329-229694442      - |             ABCB10
##               ...      ...                 ...    ... .                ...
##   ENSG00000132003    chr19   13906274-13943044      + |             ZSWIM4
##   ENSG00000162415     chr1   45482071-45771881      - |             ZSWIM5
##   ENSG00000203995     chr1   53308183-53360670      + |             ZYG11A
##   ENSG00000162378     chr1   53192126-53293014      + |             ZYG11B
##   ENSG00000036549     chr1   78028101-78149104      - |               ZZZ3
##                   ensembl_gene_id
##                       <character>
##   ENSG00000188984 ENSG00000188984
##   ENSG00000204518 ENSG00000204518
##   ENSG00000108270 ENSG00000108270
##   ENSG00000198691 ENSG00000198691
##   ENSG00000135776 ENSG00000135776
##               ...             ...
##   ENSG00000132003 ENSG00000132003
##   ENSG00000162415 ENSG00000162415
##   ENSG00000203995 ENSG00000203995
##   ENSG00000162378 ENSG00000162378
##   ENSG00000036549 ENSG00000036549
##   -------
##   seqinfo: 24 sequences from an unspecified genome; no seqlengths

Multi Assay Experiment object

A Multi Assay Experiment object from the MultiAssayExperiment package is the input for multiple main functions of ELMER. It contains the above components and making a Multi Assay Experiment object by createMAE function will keep each component consistent with each other. For example, althougth DNA methylation and gene expression matrixes have different rows (probe for DNA methylation and gene id for gene expression), the column (samples) order should be same in the two matrixes. The createMAE function will keep them consistent when it generates the Multi Assay Experiment object.

## [1] TRUE

You can also use your own data and annotations to create Multi Assay Experiment object.

Bibliography

Durinck, Steffen, and others. 2009. “Mapping Identifiers for the Integration of Genomic Datasets with the R/Bioconductor Package biomaRt.” Nature Protocols 4 (8). Nature Publishing Group:1184–91.

Ramos, Marcel, and others. 2017. “Software for the Integration of Multi-Omics Experiments in Bioconductor.” Cancer Research 77(21); e39-42.

Yates, Andrew, and others. 2015. “Ensembl 2016.” Nucleic Acids Research. Oxford Univ Press, gkv1157.

Zhou, Wanding, and others. 2016. “Comprehensive Characterization, Annotation and Innovative Use of Infinium Dna Methylation Beadchip Probes.” Nucleic Acids Research. Oxford Univ Press, gkw967.