tag:blogger.com,1999:blog-10781944.post4195009718136683422..comments2024-03-28T00:36:36.460-07:00Comments on The Tree of Life: I think that I shall never see - metagenomic analysis as lovely as a tree #PhylogenyRules #PLoSOneJonathan Eisenhttp://www.blogger.com/profile/07953790938128734305noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-10781944.post-56794718864879996762011-09-06T19:40:10.537-07:002011-09-06T19:40:10.537-07:00Note - Jack Gilbert on twitter (see here ) and J...Note - Jack Gilbert on twitter (see <a href="http://twitter.com/#!/gilbertjacka/status/109025871695855616" rel="nofollow"> here </a>) and Jed Fuhrman in email both have asked about possible issues with analyzing data from fosmid libraries as we did in the paper. <br /><br />Jed wrote in an email<br /><br />" I just saw your interesting paper in PLoS one (Kemble et al.), and noticed something that you may not be aware of, which potentially has some impact on the interpretation of results. .... Your paper notes that the SAR11/Pelagibacter clade is thought to be among the most abundant group of organisms at the site (I agree- confirmed by numerous approaches including FISH), and are abundant in the 16S libraries, yet their 16S rRNA genes are absent from the metagenomic fosmid libraries (from the top 700 m) and other SAR11 genes are relatively low in abundance in that library. The explanations mentioned in your paper are possible amplification bias and copy number differences, which may explain part of it. An alternative and highly likely explanation is that fosmid libraries (in E. coli) can be strongly biased against certain genes (by GC content and other factors) to the point they can completely exclude many clearly abundant genes. My question is then - if the 16S (and adjacent) genes of one of the most abundant organisms are completely absent, is it wise to be interpreting relative gene abundances at all? Therefore I suspect that fosmid metagenomic libraries are probably not a good choice for your analysis. It may be largely a moot point in that most current and future metagenomes are likely to be from clone-free methods, although these present their own technical challenges."<br /><br />I think this is a great point of Jed's and hopefully we will see from Steven Kembel soon what happens when you use this phylogeny driven approach to analyze other types of metagenomic data.Jonathan Eisenhttps://www.blogger.com/profile/07953790938128734305noreply@blogger.com