28 Apr 2005    mcxdeblast 1.005, 05-118

NAME

mcxdeblast - parse BLAST files and write mcxassemble input.

mcxassemble will in turn create input suitable for mcl.

SYNOPSIS

mcxdeblast file-name

mcxdeblast [--score=<b|e|r> (bit scores|e-values|normalize bit score)] [--sort=<a|o> (alphabetic|occurrence sorting)] [--m9 (expect BLAST column format)] [--abc (expect ID1 ID2 score format)] [--xi-dat=<suf> (strip <suf> from file-name)] [--xo-dat=<suf> (add <suf> to base name)] [--bcut=<val> (bit score cutoff)] [--ecut=<val> (E-value cutoff)] [--tab=<fname> (tab file)] <file-name>

DESCRIPTION

mcxdeblast parses BLAST files and writes results suitable for input in mcxassemble (which will in turn create input suitable for mcl). It is used by the pipeline script mclblastline.

If the input file is named iput, mcxdeblast will by default create files iput.hdr, iput.raw, iput.tab, iput.map, and iput.err. The hdr file contains information about the number of nodes found. It will be read by mcxassemble. The raw file contains the co-occurrence scores between different nodes (peptides), also to be read by mcxassemble. The co-occurrence scores are in terms of mcl indentiers, which are subjected to rearranging by mcxassemble in order to correctly align them with the tab file. The tab file contains bindings between mcl identifiers (after rearranging as specified in the map file) and the BLAST labels (peptide identifiers). Refer to the manual of mcxassemble for the syntax of a tab file. The err file contains a list of any errors encountered.

If the --tab=fname option is used mcxdeblast changes its behaviour. It will read the file fname, and use the bindings found therein. It will no longer output any of the hdr, map, or tab files. The hdr file should be provided by the same application that generated the specified tab file, and the map file should no longer be necessary (it might be convenient though and perhaps the future will bring a new --map option).

The --tab=fname option can be convenient e.g. if you split a BLAST job over multiple machines and want to apply mcxdeblast to each of the subresults. You do this by generating a single central tab file in advance. Then provide each instance of mcxdeblast with this tab file. Each instance will generate a raw file. The respective raw files can simply be concatenated and fed to mcxassemble using its -raw option. Additionally, a hdr file has to be specified for mcxassemble using its -hdr option. The syntax of hdr files is described in the mcxassemble manual. Currently there is usually only one variable in the hdr file, which is the number of bindings listed in the tab file. This number, in turn, should equal the total number of different peptides encountered in the BLAST input.

OPTIONS

--tab=<fname> (tab file)
   
Specify a tab file to use, rather than automatically generating one.
   
--score=<b|e> (bit scores|e-values)
   
Specify whether to use bit scores, E-values, or bit scores normalized by length of the HSP.
   
--sort=<a|o> (alphabetic|occurrence sorting)
   
Specify how to sort labels.
   
--m9 (expect BLAST column format)
   
Use this option to parse BLAST output formatted with the -m 9 option.
   
--abc (expect ID1 ID2 SCORE format)
   
Use if each input line has the format ID1 ID2 SCORE.
   
--xi-dat=<suf> (strip <suf> from <file-name>)
   
If set, the suffix <suf> is stripped from file-name in order to obtain the base name for output. If <file-name> does not end with suffix, it is assumed to be the base name, and <suf> is added in order to obtain the (real) file-name.
   
--xo-dat=<suf> (add <suf> to base name)
   
Attaches the suffix to the base name. Use this to distinguish between different parses. One might for example use --score=b --bcut=5 --xo-dat=b.
   
--bcut=<val> (bit score cutoff)
   
Scores below the cutoff value are ignored.
   
--ecut=<val> (E-value cutoff)
   
Scores below the cutoff value are ignored.

AUTHOR

Stijn van Dongen. Jason Stajich implemented the --m9 option. Abel Ureta-Vidal and Dinakarpandian Deendayal contributed helpful comments and fixes.

SEE ALSO

mclblastline, mcxassemble, mcl, clmformat.