Gap4 does not store the trace data within the gap4 database. Instead it stores the filename of the trace file. Usually the trace files are kept within the same directory as the gap4 database. If this is not the case gap4 needs to know where they are.
To make sure that gap4 can still display the traces we need to specify any alternative locations where traces may be found. The "Trace File Location" command (which is available from the gap4 Options menu) performs this task. It brings up a dialogue asking for the directory names. If there is just one directory to specify, its name should be typed in. If there are several directories to search through, they must all be typed in, separated by the colon character (":"). To include a directory name that contains a colon, use a double colon.
For example, on windows to specify two directories, use (eg)
"F::\tfiles1:G::\tfiles2
".
In addition to specifying directories, RAWDATA may also be used to indicate
that the trace files come from a variety of other sources using the
general format SOURCETYPE=path. These can be combined with directories
if desired. For example ".:/trace_cache:TAR=/traces/archived.tar
".
TAR=filename.tar
index_tar
program then the trace name will be looked up in
the index instead of sequentially scanning through the tar file. In
order to speed up accessing of traces within the tar file a command
line utility named index_tar
may be used. This produces a text
index containing the filenames held within the tar and their offsets
within it. Programs will then use this index file to provide a fast
way of accessing the trace. The syntax for index_tar
is:
index_tar
tar_filename >
tar_filename.index
. (For example "index_tar
traces.tar > traces.tar.index
".)
SFF=filename.sff
HASH=archive.hash
hash_tar
tool. Similarly 454 SFF archives may be
hashed using hash_sff
. In theory any type of archive may be
indexed as a ".hash" provided that the traces are stored
uncompressed (or compressed only using their own methods, such as with
ZTR) so that random access is possible within the archive.
The Hash file contains a precomputed binary index of all the traces
contained within it stored in such a way that random access is very
fast.
URL=url
wget
tool (not supplied as part of the
Staden Package) to fetch a given url. Anywhere that %s
occurs
within the specified url will be replaced by the trace
name. Hence, for example,
URL=http://trace.server.org/cgi-bin/lookup.pl?trace=%s
could be
used to fetch named traces from a remote site. There are plans for
such URL access to be made available via the Ensembl TraceArchive.
If the gap4 database has been opened with write-access this directory
location will be stored as a database RAWD
note
(see section Special Note Types), which is read by gap4 when
it opens the database. The demonstration data supplied with the package
includes an example database (named DEMO.0) that has a RAWD note to specify
that traces are fetched from a tar file within the same directory.
An alternative way of specifying the trace file location is by setting the
RAWDATA
environment variable. On Unix and Windows NT this is
straightforward (although system and shell specific). However on Windows 95
this may prove difficult (and at least require a reboot), so manually setting
the environment variable is no longer recommended.