Genomics 101 (cancer)

Under scientific review

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

Genomics studies the full DNA blueprint of cells: sequence variants, copy-number changes, and structural rearrangements. In oncology it drives diagnosis, targeted therapy, and resistance monitoring.

The Genomic Data Flow (Pipelines)

In oncological bioinformatics, biology is transformed into strings and matrices through a standardized pipeline:

1. Raw Sequencing (.fastq): The sequencer (e.g., Illumina) spits out millions of small DNA "reads", composed of the letters A, C, T, G and a quality score for each letter (PHRED score).
2. Alignment (.bam or .sam): Software (like BWA-MEM) takes these short reads and tries to fit them into the Human Reference Genome (GRCh38). The BAM file is essentially a massive spreadsheet saying where each read aligned on the chromosome and with what margin of error.
3. Variant Calling (.vcf or .maf): Software like GATK Mutect2 or Strelka compare the patient's tumor BAM with the patient's normal blood BAM (Tumor-Normal Paired). They generate a VCF containing only the "mutations" found in the tumor. The MAF (Mutation Annotation Format) is a tabular version of the VCF, easier for data scientists to manipulate in R or Python.

Key Metrics for Data Engineers

• Depth of Coverage (Coverage): How many times the same DNA region was read. A coverage of 30x is standard for blood, but tumors need 500x to 1000x coverage because the tumor tissue is a mixture of normal and cancer cells.
• Variant Allele Fraction (VAF): If a mutation has a VAF of 10%, it means that out of 100 reads at that exact position of the genome, 10 contained the mutation and 90 were normal.

Complex Genomic Biomarkers (Signatures)

Instead of just looking for a specific mutation (e.g., BRAF V600E), modern computational genomics calculates pan-tumor signatures:

• TMB (Tumor Mutational Burden): A total count of how many mutations exist per megabase of DNA in the tumor. Tumors with "High" TMB (e.g., > 10 mut/Mb) have so many altered proteins that the immune system can notice them, making the patient a great candidate for Immunotherapy.
• HRD (Homologous Recombination Deficiency): Instead of looking for point mutations, it measures "scars" across the entire genome (large chunks of deleted or duplicated chromosomes). A high HRD indicates that the tumor has lost the ability to repair double-strand DNA breaks, making it vulnerable to PARP Inhibitors (Synthetic Lethality).

Common assay modes

Mode	Computational Scope	Typical Clinical Use
Targeted panel	Evaluates 50 to 500 genes. Generates small BAM files (gigabytes).	Gold standard in the clinic today (e.g., FoundationOne CDx). Focuses on actionable genes with approved drugs.
Whole-exome (WES)	Sequences the ~20,000 protein-coding genes (2% of the genome). Requires ~20-50GB per patient.	Used in research and for accurate TMB calculation.
Whole-genome (WGS)	Sequences all 3 billion base pairs. Requires >100GB per patient.	Finds large structural variants, non-coding drivers, and fusions that panels miss.
ctDNA (Liquid Biopsy)	Ultra-deep sequencing of tumor DNA floating in the blood.	Monitoring of drug resistance and minimal residual disease without a physical biopsy.

From files to interpretation

Pair this overview with the workflow guide From FASTQ to variants and data access patterns in Data & APIs.

See also

• Transcriptomics 101 · Epigenomics 101
• Multi‑omics integration · Omics hub