SPEC CPU2006 Benchmark Description

Benchmark Name


Benchmark Author

Sean Eddy
HHMI Janelia Farm Research Campus
19700 Helix Drive
Ashburn VA 20147

Project Leader - Kaivalya M. Dixit (IBM)
               - Alan MacKay (IBM)

Benchmark Program General Category

Search a gene sequence database

Benchmark Description

Profile Hidden Markov Models (profile HMMs) are statistical models of multiple sequence alignments, which are used in computational biology to search for patterns in DNA sequences.

The technique is used to do sensitive database searching, using statistical descriptions of a sequence family's consensus. It is used for protein sequence analysis.

Input Description

A database (sprot41.dat) is used in the reference workloads. An input workload (nph3.hmm) is used to find a ranked list of best sorting sequences from the sprot41.dat file using the hmmsearch function.

For test, train, and 1 of the 2 reference workloads, 3 different hmm files are used to with the hmmcalibrate function to calibrate HMM search statistics. This function scores a large number of synthesized random sequences from the input file and fits an extreme value distribution (EVD) to the histogram of those scores.

Output Description

Four output files contains a ranked list of matches.

Programming Language

C Language

Known portability issues

All resolved during porting it for SPEC.


Version - 2.3

Last updated: $Date: 2011-08-16 18:23:17 -0400 (Tue, 16 Aug 2011) $