Epigenomics 101 (cancer)

Under scientific review

This introductory page is under scientific review. Use it for orientation, not clinical decision-making.

Epigenomics captures heritable-but-reversible regulation on top of DNA: methylation, histone marks, chromatin accessibility. It explains cell state, plasticity, and sometimes subtyping without a DNA sequence change.

The Data Problem (The Control Layer)

The genome of a skin cell is exactly the same as that of a neuron or a liver cell. What defines cellular identity is Epigenomics (the "software" running on top of the DNA "hardware"). In cancer, cells "forget" their identity due to epigenetic bugs, regressing to stem-cell-like states.

Oncological data engineers deal with two main strands:

1. DNA Methylation

Methyl groups ($CH_3$) are added to the letter "C" of the DNA (Cytosine). A lot of methylation in the promoter of a gene acts like a "Mute" button.

• The data challenge: The human genome has ~28 million "CpG" sites where methylation can occur.
• Arrays (Illumina EPIC 850k): The clinical/cheap standard. Measures only ~850 thousand specific points. The generated data are values called "Beta-values", ranging rigidly from 0 (unmethylated) to 1 (fully methylated).
• WGBS (Whole Genome Bisulfite Sequencing): Sequences everything. It is the definitive approach, but computationally brutal. Since the bisulfite chemistry converts unmethylated "C"s to "T"s, genome alignment becomes very complex, because the software needs to figure out if a "T" in the FASTQ file was an original "T", a mutated "C", or an unmethylated "C".

2. Chromatin Accessibility (ATAC-seq)

DNA is 2 meters long and is coiled inside a microscopic nucleus (in the form of Chromatin). For a gene to be activated, the chromatin needs to physically "open" in that region. ATAC-seq drops an enzyme (Transposase) into the nucleus that cuts the DNA only in areas that are open.

• Peak Calling: In transcriptomics, you count reads per gene. In ATAC-seq, you search mathematically for "mountains of reads" in random regions of the genome using software like MACS2. These mountains are called Peaks.
• The .bed Format: Epigenomics data are frequently saved in BED format, which is a simple spreadsheet with the columns: Chromosome | Start Point | End Point. It is the lingua franca for cross-referencing genomic data. You can open huge files visually using the UCSC Genome Browser.

The ENCODE Project

To standardize global epigenetic data, the ENCODE consortium developed strict pipelines. Today, most ChIP-seq and ATAC-seq analyses use the pipelines maintained in Nextflow or Cromwell based on the ENCODE standard to ensure that one lab's "Peak Calling" can be compared with another's.

Cancer angles

• CIMP-like methylation patterns in some tumor types
• Fusion oncoproteins and chromatin remodelers (AML, sarcoma contexts)
• Drug targets in the epigenetic machinery (writers, erasers, readers)

Relation to genomics

Drivers can be mutations in epigenetic regulators or epigenetic silencing of tumor suppressors. Integrate with DNA and RNA for interpretation; see Multi‑omics integration.

See also

• Genomics 101 · Transcriptomics 101
• Omics hub