Alignment Scoring: Substitution Matrices, Gap Penalties and Significance · seed 1 · A4, ink-friendly. The answer key prints on its own page for grown-ups.

Score it, penalise gaps, doubt the luck

Science · Genetics & Evolution · ages 22-23
Name ______________________   Date ____________

  1. What do gap penalties model?

    • Sequencing machine colours
    • Insertions and deletions, with opening costlier than extension
    • The age of the database

  2. In a log-odds matrix, which pairs get positive scores?

    • Pairs seen more often than chance in trusted alignments
    • Pairs never observed anywhere
    • Pairs picked at random

  3. An E-value of 0.001 means about one thousand chance hits are expected.

    Circle one:   True   False

  4. A hit has E-value 2. The database doubles with the same composition. What is the new E-value?

    Answer: ______________

  5. You align two very distant cousins with a strict close-relative matrix and steep gaps. What goes wrong?

    • Real homology blurs into noise
    • The alignment becomes automatically perfect
    • E-values stop depending on database size

  6. Two hits score equally, but one comes from a tiny database and one from a giant database. Which is more trustworthy?

    • The one from the giant database
    • Both are exactly equal evidence
    • The one from the tiny database

  7. If two sequences truly share ancestry, any matrix and any gap penalty will reveal it.

    Circle one:   True   False

  8. A student claims homology from a high raw score alone, ignoring matrix and E-value. What is missing?

    • Nothing, raw scores prove everything
    • Proof the sequences are short
    • Whether the score beats chance under a fitting matrix and database size
LightMySky · lightmysky.comW1-mt_k--A8iF5SC-s1

Answer key

For grown-ups. Fold this page away before handing over the rest.

Score it, penalise gaps, doubt the luck W1-mt_k--A8iF5SC-s1

  1. Insertions and deletions, with opening costlier than extension · Gaps stand for indel events, and one event can span residues.
  2. Pairs seen more often than chance in trusted alignments · Related pairs beat chance and earn positive scores.
  3. False · It means about one thousandth of one chance hit: a trustworthy match.
  4. 4 · Expected chance hits scale with database size, so doubling doubles E.
  5. Real homology blurs into noise · The wrong matrix punishes the many real changes that distance brings.
  6. The one from the tiny database · The same score surprises more where luck had fewer chances.
  7. False · Mismatched scoring buries real homology, so the choice of matrix matters.
  8. Whether the score beats chance under a fitting matrix and database size · Raw scores mean nothing until priced against chance with the right matrix.
Worksheet · LightMySky