Step 1/3 — Upload EPICv2 .idat Files (Grn & Red)
Uploading… 0%
Upload finished. Saving files…
FOR RESEARCH USE ONLY
The AI4PH classifier was trained on the 2,801 reference samples assayed on Illumina 450K arrays from the dataset of Capper et al. (2018). The classification involves 91 distinct methylation classes of central nervous system tumors, originally established by Capper et al. (2018). The feature space is reduced using Sparse Random Projection (SRP) and then classified by a Logistic Regression model with L2 regularization trained on the full reduced dataset using the SAGA optimizer and class-balanced weights. The classifier was evaluated using a 5-fold stratified cross-validation method achieving 97.24% accuracy.