Question

Which of the following is not an advantage of Statistical models’ methods in analyzing protein motifs?

(a) Sequence information is preserved from a multiple sequence alignment and expresses it with probabilistic models

(b) Statistical models allow partial matches and compensate for unobserved sequence patterns using pseudo-counts

(c) Statistical models have stronger predictive power than the regular expression based approach, even when they are derived from a limited set of sequences

(d) The comparative flexibility is less in case of these methods when compared to regular expressions methods

I had been asked this question in an online interview.

Origin of the question is Motif and Domain Databases Using Statistical Models in portion Sequence Alignment of Bioinformatics

Answer 1

Right choice is (d) The comparative flexibility is less in case of these methods when compared to regular expressions methods

The best explanation: The major limitation of regular expressions is that this method does not take into account sequence probability information about the multiple alignment from which it is modeled making them less flexible. If a regular expression is derived from an incomplete sequence set, it has less predictive power because many more sequences with the same type of motifs are not represented. Unlike regular expressions, position-specific scoring matrices (PSSMs), profiles, and HMMs preserve the sequence information from a multiple sequence alignment and express it with probabilistic models.