iqtree
Mysterious .parstree file generated #482
-
|
I reconstructed a tree for a Thioredoxin alignment with: iqtree2 -s COG0386_aligned.faa -m LG+C20+G4 -bb 1000 -bnni -wbt Everything ran fine, but there is an additional file that I had not seen before and which is not specifically described in the log as output. The file is a tree with the .parstree extension. If I search for parstree in the documentation, nothing appears. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 2 replies
-
|
Hi @microfafa-gh, this is the parsimony tree generated at the start of the analysis. I'm sure you know this, but just in case others are wondering why this is generated - phylogenetic analysis have to start somewhere. One option is to start on a random tree, but this usually means that stuff just takes a long time (most random trees are really bad!). A parsimony tree is usually very quick to calculate, and a good first approximation. The ML algorithm can then iteratively improve this through model optimisation and heuristic tree searching. Hope that helps! Rob |
Beta Was this translation helpful? Give feedback.
-
|
Hi @roblanf, thanks for your answer. I guess I'm curious why it was written on the disk and not cleaned after the analysis was done. No difference with the other files I'm running in parallel, and the final tree looks good, but it caught my attention! Some cluster glitch maybe... FLH |
Beta Was this translation helpful? Give feedback.
-
|
Not sure why it's written. I have to say I thought it was always there, but I haven't really paid attention. Maybe @bqminh can enlighten us! |
Beta Was this translation helpful? Give feedback.
Hi @microfafa-gh, this is the parsimony tree generated at the start of the analysis. I'm sure you know this, but just in case others are wondering why this is generated - phylogenetic analysis have to start somewhere. One option is to start on a random tree, but this usually means that stuff just takes a long time (most random trees are really bad!). A parsimony tree is usually very quick to calculate, and a good first approximation. The ML algorithm can then iteratively improve this through model optimisation and heuristic tree searching.
Hope that helps!
Rob