search

BDI-pathogens / phyloscanner

Phylogenetics between and within hosts at once, all along the genome.
GNU General Public License v3.0
47 stars 14 forks source link

Seperate analyses or combined dataset #41

Closed evandiego closed 5 years ago

evandiego commented 5 years ago

Hello.

I am seeiking advice for running phyloscanner. I am a looking at a cohort of sequences where I have the donor and recipient. Would you suggest that I analyze each donor and recipient together in phyloscanner i.e. as a pair or should I analyze all of the data together i.e. as in the entire cohort combined as a single dataset? Have the authors noticed a difference in the results using the two methods?

Also in order to figure out the transmission direction is there a easy way or an easy summary that summarizes which is ancestral and whether the host changes are in one direction or what the earliest node corresponds to?

Thank you,

mdhall272 commented 5 years ago

We haven't noticed a huge difference. In principle the inclusion of more samples in a single run should help with rooting. I would say that it's better to do combined runs unless that is going to be computationally prohibitive.

There are a number of outputs that will answer your second question. Are you using NGS data?

evandiego commented 5 years ago

Thank you for your reply. Yes I have NGS data. Illumina.

mdhall272 commented 5 years ago

Hi, sorry for the delay. Look at the manual regarding classification of pairwise relationships and the pairwise relationship summary in the first instance?

evandiego commented 5 years ago

Thanks Matthew for these patience. I have looked at the manual and perfromed the phyloscanner. However, most of my patients appear to have a "complex" relationship and when I go to figure out directionality I only get results from host1->host2 when I use the RScript called transmission summary.R. Is it possible to get direction when relationship is complex? How would I figure out the frequency of host2->host1 or equivocal? Does the program automatically collate the path12 and path21 over every window? Thank you.

mdhall272 commented 5 years ago

Did you run phyloscanner_analyse_trees.R with --allowMultiTrans ?

evandiego commented 5 years ago

Yes I used the allowMultiTrans option.

mdhall272 commented 5 years ago

That does suggest that you don't have a great signal of directionality. We call the "complex" topology when we aren't confident in the direction. Most likely the tips from the pair are very intermingled. I'm happy to have a look at the files if you like.