Performance Comparison

Algorithm Recall Precision Speedup
RUBBLE 96% 99% 13X
BLASTp-fast 77% 95% 11X
BLAT 19% 33% 177X
LAST 7% 67% 35,000X


For in depth docs and instructions, please see the RUBBLE main Github page.


RUBBLE has one external dependency, and it's NCBI BLAST+. Before you can run this pipeline you will need to be sure that all executables (especially blastp, makeblastdb, and blastdbcmd) are installed on your machine and included in your PATH. Latest versions of BLAST binaries are located here.

Reference Rubble

Nasko et al. Manuscript in preparation.


RUBBLE is a BLAST-based pipeline that allows you to perform BLASTp searches 10-20X faster, without compromising your results (precision = 98% [±2%]; recall = 98% [±2%]). RUBBLE speeds up BLASTp searches no matter how big or small your query dataset or subject database are. However, RUBBLE is perhaps most useful when the query dataset and subject database are very large (e.g. UniRef100 ).

RUBBLE Workflow