Download Typhon

License

Typhon is distributed under the GNU public license.

Installing Typhon

To install Typhon, first download the tarball below and unzip it into your working directory.

Typhon uses the Mandala software package for seed design. Because the licensing terms of Mandala prohibit redistributing the source code, to use Typhon you must first download the Mandala version 1.1.1 source code from their website.

As of 06/06/2005, the code was available at http://www.cse.wustl.edu/~jbuhler/mandala/mandala-1.1.1.tar.gz

Once you have obtained this package, place the file in the 'extern' directory that will be created under the Typhon root directory.

To build Typhon, first check the file System.h to ensure that the options listed therein are correct for your system. In particular, enter the class that should be used for hash tables. If your c++ implementation does not support hash tables, or if you have trouble getting this to work, then use std::map and std::multimap.

Once you have edited System.h, edit the Makefile to incorporate any system specific compiler information. When done, typing "make" in the top level directory should build typhon into the "typhon" executable. For information on running typhon, see below.

Please note that we have not tested Typhon on any platform besides Linux with gcc version 3.3 or higher as a compiler. This means that Typhon may not compile on your system, and may not run correctly if it does. We make no claims in these respects.

That said, if you encounter any difficulties or bugs we would like to do our best to correct them. Please contact us.

Beta version

Typhon is currently in beta. While almost fully functional, some (important) functionalities are planned for later versions. In particular:

Building and writing an index to disk is not supported. To run many queries against the same index, cat them together in .mfa format and use the -sepq option.
Memory issues may be a problem on machines lacking a lot of memory. On our tests the memory issues were typically about 150 bytes for every base of the sequence (i.e. 300MB for a 2Mbp sequence). However, spurts were up to 3 times that, albeit for very short amounts of time (at the very beginning of the program). Things that you can do to decrease memory usage:
- Use the -memopt option. This will most likely degrade performance.
- Change your phylogenetic tree. Different trees can have different memory requirements. The most expensive is a complete binary tree, while an almost linear tree is the least expensive. We haven't formally analyzed this, but you can fiddle around to see what happens. This will lower the initial spurt in memory usage, but will leave the asymptotic memory usage the same.
- Use a smaller pattern weight. This is why we currently don't support weights greater than 12. If you want to get fewer hits, try using the -hitdist option. We haven't tested the performance of this, however.
- Running times are okay but can be improved, particularly for building the profile and the index.
- Portability has not been determined. Please contact us as you encounter issues.
- The code is slightly messy, and should be cleaned up.
- While we have tested Typhon and have fixed every seg fault we have ever encountered, undoubtably there exist bugs of which we are unaware. Please let us know if you find any.

Download Typhon Beta

Back