Classifiers methods

Classifying gliomas samples with gliomaClassifier

Classifying glioma samples with DNA methylation array based on:

Ceccarelli, Michele, et al. “Molecular profiling reveals biologically discrete subsets and pathways of progression in diffuse glioma.” Cell 164.3 (2016): 550-563. (https://doi.org/10.1016/j.cell.2015.12.028)

Possible classifications are:

• Mesenchymal-like
• Classic-like
• G-CIMP-high
• G-CIMP-low
• LGm6-GBM
• Codel

Data

The input data can be either a Summarized Experiment object of a matrix (samples as columns, probes as rows) from the following platforms:

• HM27
• HM450
• EPIC array.

In this example we will retrieve two samples from TCGA and classify them expecting the same result as the paper.

query <- GDCquery(
    project = "TCGA-GBM",
    data.category = "DNA Methylation",
    barcode = c("TCGA-06-0122","TCGA-14-1456"),
    platform = "Illumina Human Methylation 27",
    data.type = "Methylation Beta Value"
)
GDCdownload(query)
dnam <- GDCprepare(query)

assay(dnam)[1:5,1:2]

Function

classification <- gliomaClassifier(dnam)

Results

The classfier will return a list of 3 data frames:

1. Sample final classification
2. Each model final classification
3. Each class probability of classification

names(classification)
classification$final.classification
classification$model.classifications
classification$model.probabilities

Comparing results with paper

TCGAquery_subtype("GBM") %>%
 dplyr::filter(patient %in% c("TCGA-06-0122","TCGA-14-1456")) %>%
 dplyr::select("patient","Supervised.DNA.Methylation.Cluster")

## gbm subtype information from:doi:10.1016/j.cell.2015.12.028

##        patient Supervised.DNA.Methylation.Cluster
## 1 TCGA-06-0122                       Classic-like
## 2 TCGA-14-1456                        G-CIMP-high