Bioinformatics

Aligning two DNA or protein sequences finds the best mapping of one to the other, allowing substitutions, insertions and deletions. The Needleman-Wunsch algorithm does global alignment; Smith-Waterman does local. Scoring matrices (PAM, BLOSUM) capture biological likelihoods.

CRISPR-Cas9 uses a guide RNA matching ~20 bases of target DNA. Bioinformatics tools search the genome for: (1) target sites flanked by PAM motifs (NGG for Cas9), (2) off-target hits using mismatch tolerance, (3) gRNA efficiency scores from machine learning.

A phylogenetic tree visualises evolutionary relationships. Distances are inferred from sequence differences (number of mutations between species). Methods: neighbour-joining, UPGMA (clustering), maximum likelihood. The tree topology hypothesises the order of speciation events.

Basic Local Alignment Search Tool — finds similar sequences in a database faster than full alignment. It hashes short 'words' (3-letter for proteins, 11-letter for DNA), seeds matches, then extends them. The E-value rates statistical significance.

Proteins fold from linear amino acid chains into 3D structures that determine function. AlphaFold and ESMFold use deep learning to predict structure from sequence. The energy landscape (Anfinsen's hypothesis): the native fold sits at the global energy minimum, found via funnel-like landscapes.