We now have GFS running on Windows XP. It will be available for download from our website as a Windows executable file shortly. It seems to run a bit more slowly on Windows than under Linux or OSX on the same processor. We haven't yet determined the source of this, though we suspect that it has to do with memory allocation speed when using the Gnustep kit under Windows. However, we've done a lot of overall speed optimization lately, so hopefully it won't be too slow... we welcome user feedback and input, especially if you have insights on this.

In other news, we are considering moving the GFS project over to sourceforge. We welcome your feedback about this (remove "remove.these.words." from the email address).

The front page of the Journal BMC Evolutionary Biology features our paper on the evolution of the Dscam gene family in their Research Highlights section on the front page.

This is gratifying, since this paper has been a long time coming. Among other things, the lead author, Mack Crayton, was displaced from his new home in New Orleans while we were in the revision process.

This paper has now been rated as "highly accessed" by BMC Evolutionary biology.

For those wondering what's going on with GFS these days, here's a rundown of some highlights (more detail coming in a paper soon);

- We have GFS running on Debian Linux now, and it seems to run just fine. It relies on the GnuSTEP object kit, which is essentially a replica of Cocoa in MacOSX. We rely on this object kit because it provides extensive functionality such as that which is now present in Java, e.g. lists, arrays, strings, sets, and all sorts of useful stuff. We are now working on testing and packaging GFS for Linux. Next we will work on the Windows port, which is not expected to be difficult, since GnuSTEP is also available for Windows, and the port to GnuSTEP is done.

- We have added MS/MS interpretation via sequence tags. Presently it only works with combined MS and MS/MS data (i.e. peptide mass fingerprints). However, we are extending the program to operate with MS/MS spectra alone, e.g. for MudPIT type data.

- GFS is now designed for automatic distribution of large jobs on a compute cluster. Presently the functionality is designed for Sun Grid Engine, but can be easily extended to other queue management software.

- We now have automated the calculation of the Expectation value for each match made, using the approach originally developed by Fenyö.

- Many optimizations have been made to speed searches and improve the clarity of the results. And, of course, there's the occasional bug fix as well.