Bioinformatics & Application
From raw reads to a phylogeny — the pipeline, and where it lies to you.
About this section
Section 2 covered methods that learn from data. This section covers the computational pipeline that produces that data — and the artifacts it introduces at every step.
The running example is hepatitis B virus, using real sequencing data from Botswana.
The modules
| Module | Core idea | |
|---|---|---|
| 1 | Sequence Data & Quality Control | FASTQ, Phred scores, adapters; what to trim and what to keep |
| 2 | Mapping & Variant Calling | Reads to reference; why coverage and depth are different |
| 3 | Genotyping & Recombination | Assigning genotype; why recombinants break naive classifiers |
| 4 | Phylogenetics | Building trees; what bootstrap values do and don’t mean |
The through-line
Extraction favours some templates. PCR duplicates inflate apparent depth. Basecalling guesses. Alignment forces a read to a position that may be wrong. Variant calling applies a threshold someone chose.
By the time you have a VCF, five judgement calls stand between you and the biology. Knowing they happened is the difference between analysis and credulity.
Prerequisites
Comfort with the Unix command line. The pipeline tools run outside R.
Module content is under active development for this workshop. Structure and learning objectives are settled; worked examples are being finalised against the HBV sequencing datasets.