How sirselim/illumina450k_filtering works
Project type and purpose
• This is an R‐based bioinformatics QC/filter pipeline, not a full‐stack application.
• Purpose: remove ‘bad’ (cross-reactive, multimapping, variant‐containing, or detection-failed) probes from Illumina 450k/EPIC methylation array data prior to downstream analysis.
File structure
• Top level:
• CSV/TXT resources for known ‘bad’ probe lists
• EPIC/ folder holding six supplementary CSVs (cross-reactive & variant lists)
• A single README with usage instructions and R code snippets
• No backend/frontend division—everything runs inside R.
• No build or deployment scripts; it’s a data resource + R pipeline.
README for architectural clues
• Workflow stages:
a. Probe sequences (Bowtie2) → generate multimapping list
b. Import non-specific probe list (Chen et al.)
c. Detect failed probes in samples via detectionP()
d. Import EPIC lists (Pidsley et al.)
e. Combine all lists → filter β‐value matrix
• Dependencies: R, Bioconductor (minfi), Bowtie2 outputs, Illumina manifest.