Hi, great question and a powerful approach. My first thought is to consider the traits separately, # of hosts and virulence. Then consider SFP haplotypes throughout the genome in sliding windows, but blocks would be preferred. You need to associate a plasmodium haplotypes with the quantitative trait across strains and then scan the genome. Your null can be shuffled phenotypes vs true haplotypes to determine a genome wide FDR. I've been measuring pairwise diversity and total SFP counts (and their difference Tajimas D like) thinking about patterns of selection on the genome but this measure isn't what you want for an association. It could however give some clues as to what might be going on. I'm sure you've seen recent papers by Kim et al Kidgell et al. Good luck! (the more diverse strains with good tests of host range/virulance the better >100) Justin ----- Justin Borevitz http://naturalsystems.org/lab -----Original Message----- From: Charles C. Berry [mailto:cberry@tajo.ucsd.edu] Sent: Monday, April 09, 2007 10:18 AM To: Justin Borevitz Subject: [BioC] what is the best way to correlate SFP data to a phylogenetic correlation? (fwd) Justin, Did you see this? I did not see any traffic on bioc in response. Best, 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 ---------- Forwarded message ---------- Date: Thu, 5 Apr 2007 19:20:29 -0700 (PDT) From: lgilbert@berkeley.edu To: bioconductor at stat.math.ethz.ch Subject: [BioC] what is the best way to correlate SFP data to a phylogenetic correlation? I don't know where to start: I'm planning on generating a single feature polymorphism (SFP) map for 11 primate strains of malaria using a high density array. Some previous work on these primate malarias found a significant negative correlation between a quantitative measure of virulence and the number of host species each primate malaria could infect using a least squares model. I want to somehow test or establish an association with the SFP divergence score for genes implicated in host specificity to this other data- the host number-virulence correlation. I don't know what would be the best statistical approach- some likelihood method, a multivariate correlation? Sincerely, Betty