crystalmeth: crystalmeth: A streamlined workflow for DNA methylation-based...

In cgeisenberger/crystalmeth: Streamlined Methylation-Based Classification of Tumors

Description

The crystalmeth package provides a coherent framework to use DNA methylation data for diagnostic purposes. Essentially, the workflow is split into four step which are outlined in further detail below. For more information and feature requests, refer to the Github Repository.

Details

The major steps in processing samples are:

1. Detect input (via scan_directory())

2. Initiate an object of class ClassificationCase for every pair of IDAT files

3. Process samples

4. Render a diagnostic report for each case

Input detection

The first step is to load data for the cases that should be classified. Currently, only raw data (IDAT files) are supported as some features (such as CNV calling) need access to raw intensity measurements of the array. Support for loading other types of data suchs as Beta values, MSet or RgSet objects may be included in future versions.

Sample processing

After detecting the input files, data has to be loaded into memory. Crystalmeth offers a handy object type called ClassificationCase for this purpose. Upon initiation, it performs a thorough check of the input files. Afterwards, a single command case$run_workflow() will load, preprocess and impute missing data. Furthermore, the sample is classified, tumor purity estimation performed and a copy-number profile is generated.

Report generation

The package comes with the function render_report(), a handy wrapper around rmarkdown::render(). It will create a report based on a markdown template for an object of type ClassificationCase.

cgeisenberger/crystalmeth documentation built on Aug. 28, 2020, 9:35 a.m.