Question 2 of 6 Fundamentals Interpretation
How should E-values be interpreted in database sequence searches?
Short answer
A BLAST E-value estimates the number of alignments with a comparable or better score expected by chance under the search conditions. It is not a probability that a hit is biologically false.
Why
An E-value reflects an extreme-value model for random sequence matches, with effective database size, query length and scoring statistics affecting the expectation. In a particular database, E = 0.01 roughly means 0.01 matches at least this strong are expected by chance under the model. The value can change if you search a much bigger database without changing the query.
Compare alignments with identity, coverage, low-complexity masking, query composition, organism and domain architecture. Extremely short identical motifs can be biologically uninformative despite impressive-looking percent identity. When screening candidate homologs, retain the search parameters and database snapshot so E-values are interpretable.
Statistical significance alone does not guarantee orthology, correct annotation or the same biochemical function.
Common mistakes
- Explaining E = 0.01 as a one-percent probability that the alignment is wrong
The interviewer might ask next
Why could the same pair of sequences receive a different E-value in a bigger database?