HMMER User's Guide

| Dept. of Genetics | WashU | Medical School | Sequencing Center | CGM | IBC| | Eddy lab | Internal (lab only) | HMMER | PFAM | tRNAscan-SE | Software | Publications |

Environment variables

HMMER is built to coexist peacefully with the BLAST suite of database search programs [Altschul, 1991]. HMMER reads the following environment variables (the examples given use UNIX csh syntax):

• [BLASTDB] Location of sequence databases that hmmsearch will look in, in addition to the current working directory. Multiple directories are allowed, separated by colons. A trailing slash on each path is important to BLAST, but not to HMMER.
  Examples:
  > setenv BLASTDB /nfs/databases/
  > setenv BLASTDB /nfs/databases/:/nfs/moredatabases/
  

• [BLASTMAT] Location of substitution matrices that hmmbuild -pam (the PAM prior option) can read. Although HMMER can parse a colon-separated list, BLAST must have a single directory path here. Example:
  > setenv BLASTMAT /nfs/databases/matrix/
  

• [HMMERDB] Location of HMMs, PFAM, or other HMMER specific data files. Any program that reads an HMM file looks in both HMMERDB and the current working directory. Multiple directories are allowed, colon-separated. Examples:
  > setenv HMMERDB /usr/local/lib/hmmer/
  > setenv HMMERDB /usr/local/lib/hmmer/:/nfs/databases/pfam/
  

• [HMMER_NCPU] On multiprocessors that support POSIX threads (this includes almost all modern UNIX multiprocessors; for example, SGI Origin servers), the programs hmmcalibrate, hmmpfam, and hmmsearch run as parallelized, multithreaded applications. Normally they will take over all available CPUs in the machine. HMMER_NCPU sets a maximum number of CPUs to utilize, so HMMER searches are ``good citizens'', leaving some CPU power for other jobs. An example of configuring HMMER to use only 16 processors on a 32-processor Origin:
  > setenv HMMER_NCPU 16