Question 5 of 6 Advanced Scenario
When does database size affect the significance of a BLAST hit?
Short answer
The expected number of random high-scoring matches rises with effective search space. As a database becomes larger, the same alignment score can produce a less significant E-value.
Why
BLAST assesses an alignment relative to how many possible chance matches are available under its statistical model. If the database grows tenfold and the score remains fixed, the expected number of random alignments can rise correspondingly, with important corrections for effective lengths and composition. In an interview, distinguish a raw or bit score, which summarizes alignment strength, from the E-value, which also depends on search space.
Re-running an old query against a newer database can change which hits pass a fixed E-value threshold. Save the database release, BLAST program, filtering settings and query sequence for reproducibility. Do not infer that a homology relationship itself has changed simply because the reported statistical expectation changed.
Common mistakes
- Saying the same BLAST score must always correspond to the same E-value regardless of the database
The interviewer might ask next
Would you compare E-values across runs using different database sizes without further context?