From a gene list to biology: enrichment tests and their backgrounds · seed 1 · A4, ink-friendly. The answer key prints on its own page for grown-ups.

Testing enrichment against an eligible background

Science · Genetics & Evolution · ages 23-24
Name ______________________   Date ____________

  1. For cutoff-based gene selection, which background matches the selection process?

    • Every genomic gene, including those never measured
    • All genes eligible to enter the selected list
    • Only selected genes after the significance cutoff

  2. What does an over-representation test assess for a selected gene list?

    • An unusual category count relative to its background
    • Experimental confirmation of every selected gene's function
    • Whether the assay measured every genomic DNA base correctly

  3. In read-count RNA sequencing, longer transcripts tend to yield more reads at equal transcript abundance.

    Circle one:   True   False

  4. A read-count analysis selects many long, highly expressed genes. What addresses possible selection bias without assuming the biology is absent?

    • Check selection chances and use a suitable bias-aware test
    • Delete all long genes and declare the remaining terms confirmed
    • Report every selected category as proof of pathway activation

  5. There are 20 eligible genes, including 5 category members. Draw 2 without replacement with equal chances; both are category members. What is the one-sided p-value for 2 or more members?

    • 1/4, about 0.250
    • 1/2, about 0.500
    • 1/19, about 0.053

  6. Eligible genes rank P1, A, P2, P3, B, C. P genes share a pathway; only P1 passes the shortlist cutoff. What extra information can a rank-based test use?

    • Every ranked gene has already passed the shortlist cutoff
    • P2 and P3 also occur toward the top of the ranking
    • The ranking proves the pathway is causally activated

  7. After testing 1,000 gene categories, an unadjusted p-value of 0.01 alone establishes a result corrected for multiple testing.

    Circle one:   True   False

  8. A term is significant with an ineligible background but non-significant with the eligible background. A rank-based test is also non-significant. What is justified?

    • The first p-value proves activation despite its ineligible background
    • The later tests establish that the pathway has no biological effect
    • The first claim lacks support; later tests do not prove no effect
LightMySky · lightmysky.comW1-mt_M9eHEXq6iA-s1

Answer key

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

Testing enrichment against an eligible background W1-mt_M9eHEXq6iA-s1

  1. All genes eligible to enter the selected list · The background represents genes that could have entered the list, not only genes that did.
  2. An unusual category count relative to its background · The test compares the observed category count with counts expected under its background-based selection model.
  3. True · Longer transcripts offer more fragment positions. This can affect detection power without making a gene's change false.
  4. Check selection chances and use a suitable bias-aware test · Length and expression can affect detection power. Examine whether unequal selection chances affect the result and model them when needed.
  5. 1/19, about 0.053 · There are 190 background pairs and 10 category-only pairs. Two is the maximum count, so the upper-tail probability is 10/190 = 1/19.
  6. P2 and P3 also occur toward the top of the ranking · Ranking retains the positions of pathway genes below the shortlist cutoff. Determining whether this pattern is significant requires the test's null model.
  7. False · An unadjusted p-value is not a multiple-testing correction. The analysis must account for the category tests and report the procedure used.
  8. The first claim lacks support; later tests do not prove no effect · The ineligible background undermines the intended inference. Later non-significance limits support without proving biological absence.
Worksheet · LightMySky