gmap_setup(1) create a genome database for GMAP or GSNAP


gmap_setup -dgenomename [-Ddestdir] [-oMakefile] FASTA


genome name
destination directory for installation (defaults to gmapdb directory specified at configure time)
name of output Makefile (default is "Makefile.<genome>")
use coordinates from an .md file (e.g., file from NCBI)
try to parse chromosomal coordinates from each FASTA header
interpret argument as a command, instead of a list of FASTA files
order chromosomes in numeric/alphabetic order (0 = no, 1 = yes (default))

Advanced options

write some output directly to file, instead of using RAM (use only if RAM is limited)
GMAP indexing interval (default: 3 nt)
PMAP indexing interval (default: 6 aa)


If you want to treat each FASTA entry as a separate chromosome (either because it is in fact an entire chromosome or because you have contigs without any chromosomal information), you can simply call gmap_setup like this:

gmap_setup -d <genome> <fasta_file>...

The accession of each FASTA header (the word following each ">") will be the name of each chromosome. GMAP can handle an unlimited number of "chromosomes", with arbitrarily long names. In this way, GMAP could be used as a general search program for near-identity matches against a FASTA file.

-M and -C
If your sequences represent contigs that have mapping information to specific chromosomal regions, then you can have gmap_setup try to read each header to determine its chromosomal region (the -C flag) or read an .md file that contains information about chromosomal regions (the -M flag). The .md files are often provided in NCBI releases, but since the formats change often, gmap_setup will prompt you to make sure it parses it correctly.
If you need to pre-process the FASTA files before using these programs, perhaps because they are compressed or because you need to insert chromosomal information in the header lines, you can specify a command instead of multiple fasta_files, like these examples:

 gmap_setup -d <genome> -E 'gunzip -c genomefiles.gz'
 gmap_setup -d <genome> -E 'cat *.fa | ./'

The gmap_setup process works best if you have a computer with enough RAM to hold the entire genome (e.g., 3 gigabytes for a human- or mouse-sized genome). Since the resulting genome files work across all machine architectures, you can find any machine with sufficient RAM to build the genome files and then transfer the files to another machine. (GMAP itself runs fine on machines with limited RAM.) If you cannot find any machine with sufficient RAM for gmap_setup, you can run the program with the -W flag to write the files directly, but this can be very slow.
-q and -Q
If you specify a smaller interval (for example, 3 for the GMAP interval), you can create a higher-resolution database, which can be useful for mapping small oligomers (smaller than 18 nt). However, the corresponding genome index files will be larger (twice as big if you specify -q 3). These index files may exceed the 2 gigabyte file offset limit on some computers, and will therefore fail to work on those computers.


Thomas D. Wu and Colin K. Watanabe


Report bugs to Thomas Wu <[email protected]>.


Copyright 2005 Genentech, Inc. All rights reserved.