Lecture 1.3

Sequence similarity

Learning objectives

After this lecture, you should be able to do the following.

Read a pairwise alignment column by column, and explain what an inserted or deleted letter does to every position after it.

  1. Identify matches, mismatches, and gaps in a written pairwise alignment.
  2. Explain why a single insertion or deletion shifts every later position, so comparing two sequences position by position misses their relationship.

Determine which of two alignments of one sequence pair scores higher under a stated scheme, and how a change to the scheme changes the winner.

  1. Given the scores of two alignments under one scheme and a change to the gap or the mismatch penalty, determine the new scores and whether the higher-scoring alignment changes.
  2. Given a match, mismatch, and gap scheme and two alignments of one sequence pair, determine the score of each and which of the two scores higher.

Interpret a dot plot of two sequences, and explain how a window filter separates real diagonals from chance matches.

  1. Explain why a lone dot-plot match is usually chance.
  2. Explain how a sliding window with an identity threshold keeps real diagonals while dropping noise.
  3. Read a dot plot, taking a diagonal as a conserved stretch, a break-and-shift as an indel.

Separate what an alignment measures from what it lets you infer about ancestry, and classify a homologous pair by the event that produced it.

  1. Contrast orthologs with paralogs, naming the event that produced each pair and what it implies about their freedom to change function.
  2. Contrast homology with similarity for a pair of sequences.

Readings

Supplemental

DTW ch. 6, §6.1
Deonier, Tavare, and Waterman frame the biology: homology versus similarity, orthologs versus paralogs, and why we align at all. Fix the vocabulary here. Similarity is measured, homology is inferred, and similarity can arise by chance, so it does not prove homology. That gap is {lo-homology-vs-similarity}. Its scoring sections repeat DEKM. One treatment is enough.
(T 1, 8, 9)
LAM ch. 5, the dotplot
Dotplots as a picture of where two sequences match, drawn before any matrix is filled. Use a dotplot to spot where similarity sits inside two sequences, and to read repeats and inversions as off-diagonal lines. That is {lo-read-dot-plot}.
(T 5, 6, 7)
Back to top