On Wednesday 21 February 2007 01:59, Charles C. Berry wrote: > On Tue, 20 Feb 2007, Dr_Gyorffy_Balazs wrote: > > Dear All! > > > > I have microarray data for cancer patients and I want to correlate the > > genes with survival. One option is the survival analysis of the PAM > > package. However, only the windows EXCEL version offers this option, but > > the package for R not. Moreover, the EXCEL version breaks down, so that I > > can see the successfull classification but I can not see which genes are > > included in the discriminative signature. Unfortunatelly, the authors of > > PAM do not respond to my questions. > > > > Questions: > > 1. is there an R version of PAM which I am unaware of? > > 2. is there other pacakge / software I can use to correlate microarray > > data and patient survival data? > > Have you looked at "Semi-Supervised Methods to Predict Patient Survival > from Gene Expression Data" > > http://dx.doi.org/10.1371/journal.pbio.0020108 > I do not want to start a flame war about methods here, but I found that approach confusing, and the R package very hard to understand; for instance, after struggling with it, I never really got what was the recommended approach from start to finish. (Note these authors have another related paper in JASA in 2006). If you do a search, you'll find there are a bunch of recente proposals for this type of problem (I think around 6 to 10 in the last 2 years). I personally find the threshold gradient descent method of Gui & Li very interesting, and a recent, modified and improved version by Ma and Huang seems to do even better. But beware that there is esentially nothing like a comprehensive, serious, exhaustive comparison of methods. (I have a partial short list of methods and references in an ms in review. I can send you that off-line if you want). Finally, you might want to check BRB Tools from Richard Simon and collaborators at the NCI. They have implemented something similar (though not identical) to the approach in Bair & Tibshirani. I do not use windoze, but if I had to use something like this, I'd rather borrow a windoze machine with BRB tools, or install an emulator, or similar before going for a deep scuba diving adventure in the Bair & Tibshirani code. Again, this might just be a consequence of my own incompetence. YMMV. <<now, the="" piece="" of="" self-advertisment="">> I have implemented the TGD approach cleaning the original Gui & Li code somewhat and parallelizing it, so it achieves speed improvements of factors of up to 100x in a cluster with 30 computing nodes (120 cores). We also have a web based application, signs (http://signs.bioinfo.cnio.es), that implements the approach in Dave et al. (a NEJM paper with Richard Simon). That might do some of what you want. All the code is freely available under a GNU GPL or an Affero GPL license (depends on the exact part) from the repositories (https://launchpad.net/signs). This is part of the ms in review that I could send you off-line. HTH, R. > ??? > > The R code Bair and Tibs used is in the supplements. > > HTH, > > Chuck > > > > Charles C. Berry (858) 534-2098 > Dept of Family/Preventive > Medicine E mailto:cberry at tajo.ucsd.edu UC San Diego > http://biostat.ucsd.edu/~cberry/ La Jolla, San Diego 92093-0901 > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor -- Ram?n D?az-Uriarte Statistical Computing Team Centro Nacional de Investigaciones Oncol?gicas (CNIO) (Spanish National Cancer Center) Melchor Fern?ndez Almagro, 3 28029 Madrid (Spain) Fax: +-34-91-224-6972 Phone: +-34-91-224-6900 http://ligarto.org/rdiaz PGP KeyID: 0xE89B3462 (http://ligarto.org/rdiaz/0xE89B3462.asc) **NOTA DE CONFIDENCIALIDAD** Este correo electr?nico, y en s...{{dropped}}