A Fast Approximate Covariance-Model-Based Database Search Method for Non-coding RNA

摘要

A covariance-model-based search method for non-coding RNA genes is proposed which is much faster than dynamic programming, but which is shown to be very effective in experimental tests. The method incorporates secondary structure information in the entire first pass of the database, unlike the usual primary-sequence-only pre-filters applied when using dynamic programming. An iterative alignment refining algorithm which starts at an ungapped alignment and successively selects alignment breakpoints gives only an approximation to the optimal alignment, but appears to be sufficient for gene localization.