Open mkellom opened 6 years ago
It is an internal error that should not happen. Are you using ModelTest-NG v0.1.3? If that is the case, I would probably need to see your data to check what is the cause.
Using v0.1.2. I'll give v0.1.3 a try and report back
There is also one possibility that have not yet been released: If your alignment has a very low proportion of invariable sites (say 0.02 or less), the optimization of +I or +I+G models might crash. I fixed that already but it is still not released. Nonetheless, you can disable +I and +I+G models with "-h ug" argument, as it does not make much sense to test for invariant sites proportion with such a low empirical rate.
Thanks, so I have a couple things to try. I'll update you with what I learn
I was unable to try out v0.1.3 because my university HPC's kernel is too old. But amending the option to "-h ug" did the trick on v0.1.2. I originally was running it with "-h uigf"
Hi Guys,
Get the same error though I am using v 0.1.6. (I guess. The --version prompt and using the software as well gives me: modeltest x.y.z. But it is installed under a folder called modeltest-0.1.6.) About the analyzed data: Gaps: 36.46 % Invariant sites: 6.28 %
@ddarriba can you please help? Got the same error message a few days ago again.
Not sure what this error means: modeltest: algo_search.c:925: pllmod_algo_spr_round: Assertion `fabs(loglh - best_lh) < 1e-6' failed.
modeltest aborts after this