Moore Notes 12 16 08
From OpenWetWareJump to navigationJump to search
Group Skype Chat
- How to build phylogenetic trees for marker gene families with complete genomes plus metagenomic reads?
- KP emailed Bret Larget (waiting response)
- SR emailed Tandy Warnow (waiting response)
- SK emailed Karen Cranston (Sanderson postdoc). She was unsure how we would actually use supermatrix methods on our data set since we don't have any information on taxonomic identity of the fragments.
- SAP (Huelsenbeck Lab)
- We could throw everything (reference alignment + metagenomic reads aligned by AMPHORA) into a regular phylogeny building program such as raxml, mrbayes, etc.
- Wiens has looked at effects of missing data on phylogenetic analyses, he generally finds that as long as there are many characters (1000bp?), individual sequences can be missing data for 90% of sites and still be placed on the tree with ~90% accuracy. i.e. see this paper.
- Our situation is similar to EST studies where they have very gappy alignments, Hartmann and Vision found that phylogenetic methods performed fairly well with a masked alignment including numerous gaps and missing data.
- Need to do simulations: many of our sequences are mostly missing
- Supertree methods and supermatrix methods http://www.citeulike.org/group/6072/article/1011645
- Jenna talked to Rick Baker: If you can't assign a taxa name to each read, then you are stuck.
- How to choose the moving window size?
- How to combine trees from different gene families to say something about diversity overall?
- Develop a method to sum/average across marker gene families
- Try to make a single tree - hard!
- JG has a simple version of this problem in another study (with 16S and 1 protein coding gene)
- Idea: make a landscape summary of phylogenetic methods for making a tree
- Look online to see what people have done to summarize the literature already
- This discussion will continue on a project page here: Phylogenetic Methods
Green notes here: File:Green 12.16 iSEEM notes.docx