Skip to the content ForoHackers
Log inJoin

Bioinformatics: start here

Bioinformatics
#3245 · · Quote · Report

Bioinformatics is the computing of biological data, above all sequences: DNA, RNA and proteins. Sequencing machines produce billions of short reads, and software has to assemble, align, compare and annotate them to find variants, genes and the evolutionary relations between organisms. It is part data engineering, part statistics, and part biology.

Key concepts

  • Sequences and formats: FASTA for sequences, FASTQ for reads with quality scores, SAM and BAM for alignments, VCF for variants.
  • Alignment: pairwise methods such as dynamic programming, and fast seeding and indexing for mapping millions of reads.
  • Assembly: rebuilding a genome from overlapping reads, with graphs as the usual structure.
  • Variants and expression: finding differences from a reference, and counting how active each gene is.
  • Phylogenetics: trees that describe how sequences, species or populations are related.
  • Pipelines: chained, versioned steps that make an analysis repeatable.

Vocabulary

  • Read: a short fragment of sequence from a sequencer.
  • Reference genome: a standard sequence that reads are compared against.
  • Annotation: the labels on a genome saying where genes and other features lie.
  • Multiple testing: the rise in false positives when thousands of hypotheses are tested together.

What to study, in order

  1. Basic molecular biology: DNA, transcription, translation, mutation.
  2. Python or R, plus the command line for handling large text files.
  3. Implement global and local alignment with dynamic programming, and score some sequences.
  4. Run a small public dataset through a standard read-mapping and variant-calling pipeline, inspecting each file.
  5. Put the steps in a workflow manager with pinned tool versions, and compute a statistic with a proper multiple-testing correction.

Related tags

#computational-biology #algorithms #data-science #shell #research-software

You are not logged in, so this goes up anonymously: no account, no name and no address is stored with it, and it may carry no link and no image. Log in or join to post under your name. Our web server's logs keep the address and time of every request for up to 14 days; to keep yours out of them, use Tor. To reach the staff in confidence, use the drop. Sign as a returning source (a key kept only in this browser)

Sending takes a few seconds here: your browser first works out a proof of work against spam.

Reply