HMMER 2.4i release notes http://hmmer.janelia.org/ SRE, Tue Dec 5 10:03:27 2006 ________________________________________________________________ 2.4i is an interim version, containing a few feature requests and bug fixes for aficionados. This is not a production release. The current production release is still version 2.3.2 (October 2003). =============== = Expostulation =============== HMMER is currently being rewritten for a planned 3.0 release. However, no end to that project is clearly in sight. In order to be able to release minor feature requests and bug fixes, I have decided to continue to develop and release the 2.x code, independent of the major changes going on in HMMER3. 2.4i ("2.4 interim") is the first release from this branch. This release should be considered to be development code. Use it if its added features are useful to you, but be cautious, as this release has not been tested as extensively as 2.3.2, the current stable HMMER release. I chose not to base this release on the 2.3.2 production code because there were several useful changes in the development code from Oct 2003 to early 2005 that made it easier to incorporate the "fragment tolerance" feature in hmmbuild. I could not use any dev code from later than spring 2005, when 3.0 development started and the development trunk became unusable. Thus, 2.4i consists of a snapshot of HMMER's development trunk as of January 2005, plus a limited merge of some later changes. Thus it does contain some experimental and unfinished code - precisely the code that led me in early 2005 to decide a complete 3.0 rewrite was necessary. That unfinished code makes me worry a bit about possible stability issues in 2.4i. I think I closed all the loose ends, but can't be sure. Please report any problems you see. ======================= = New features included ======================= -:- hmmbuild now deals with alignments that contain sequence fragments. Alignments like this have become common because of metagenomic analysis projects. (Requested by many, including Gerard Manning [Salk Institute] and Shibu Yooseph [Venter Institute].) -:- 'hmmindex --av' indexes a Pfam database better, or any database that uses . style accession numbers. The --av option tells hmmindex to index both the full accession name (that is, .) and the base accession name (that is, ) as retrieval keys. This lets you use hmmfetch to retrieve models by their base Pfam accession, even when you don't know the latest version number. (Requested by Ewan Birney, Dec 2006.) -:- A new implementation of the Altivec optimized code for PowerPC chips is included from Erik Lindahl [Stanford]. (It is activated by ./configure --enable-altivec, when you install on a Mac or IBM PowerPC platform.) This version is about twice as fast as the previous version. ============================= = New features *not* included ============================= A number of important features have been contributed to HMMER in the past three years, including optimizations for several platforms from Intel, IBM, and others; a general new framework to support many such optimizations from Jeremy Buhler [WashU]; new parallel implementations for MPI, OpenMP, and vector Crays; a BLAST-like accelerated algorithm from Elon Portugaly and Steve Johnson (HMMERHEAD); and an iterative search version (JACKHMMER) from Steve Johnson. I hate to delay the release of all these great features, but currently my plan is to release them all with 3.0 in one big bang. However, now that we have an interim 2.4 release branch where this stuff can go while 3.0 develops, waiting is not the only option. Feel free to pressure me into incorporating your favorite feature into 2.4 if you like. I'm trying to strike a balance of where I spend my time. Left to my own devices, I'm trying to invest all my time in getting 3.0 done right now. - Sean Eddy - HHMI Janelia Farm - http://selab.janelia.org/