README.md

In translational-genomics-laboratory/methXGB: Interpretation of DNA methylation data from tumour samples using XGBoost models

methXGB (pre-release version)

Interpretation of DNA methylation data from tumour samples using XGBoost models. The methXGB package contains a collection of trained models and wrapper functions that facilitate their use.

Package Installation

library(devtools)
install_github("translational-genomics-laboratory/methXGB")

Model overview

| Model family | No. models | Model class | Training dataset | Data type (features) | Dependent variable (target) | | -------------- | ---------- | ------------------ | ------------------------- | ---------------------------------------- | --------------------------- | | immunescore | 20 | XGBoost_reg:linear | TCGA various tumour types | M-values from 450k DNA methylation array | ESTIMATE.immunescore array | | purity | 20 | XGBoost_reg:linear | TCGA various tumour types | M-values from 450k DNA methylation array | InfiniumPurify |

Function overview

• The function inferMethXGB() is used to infer biological parameters in DNA methylation data by using one of the trained XGBoost models included in the methXGB package. For example, the command inferMethXGB(mval=mval_demo, model.family="purity",dataset="GBMLGG") is used to infer tumour purity using DNA methylation from the example mval_demo data frame containing M-values.
• The idatParse() functon extracts M-values from idat files (450k and EPIC platforms). It provides readily access to data in M-values format, but it is not a replacement for cumstom-built workflows.
• getModelInfo() returns the description of the models that are included in the methXGB package.

Credits

Developed by Alberto León at the PM-OICR Translational Genomics Laboratory

translational-genomics-laboratory/methXGB documentation built on May 23, 2019, 5:04 a.m.