ERROR: [FREQS] Worse log-likelihood after optimization!

[Miao Sun]

Hi,

I have been using RAxML-NG v. 0.9.0git released on 24.05.2019 on cluster. I keep running error like "ERROR: [FREQS] Worse log-likelihood after optimization":

ERROR: [FREQS] Worse log-likelihood after optimization!

Old: -3.992416

New: -3.992416

NOTE: You can disable this check with '--force model_lh_impr'

I have added option "--force model_lh_impr"  to my running cmd, but still run into the same error, and "--force model_lh_impr" seens not be recognized. 

I have also tried suggestions posted here(https://groups.google.com/forum/#!topic/raxml/gluJCb9PGvA), no good.

Please help, and let me know if you need more info

Thanks!

##############Here is my alignment and running cmd

raxml-ng-mpi --all --msa g6886_Fagus.raxml.rba --tip-inner --force model_lh_impr --prefix g6886_Fagus --threads 1 -seed 25108 --tree rand{25},pars{25} --bs-metric fbp,tbe

[Alexey Kozlov]

Hi Miao,

I just looked at your alignment and it contains just 3 alignment columns, 2 of which are invariant.
Was it intended or maybe due to an upstream error in your pipeline? Either way, you cannot expect to
infer anything meaningful from this data using ML (or probably any method), it's simply not enough
information.

I will have a look at the actual error a bit later, it seems like raxml-ng runs into numerical
issues under these extreme conditions.

Best
Alexey

> --
> You received this message because you are subscribed to the Google Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to
> raxml+un...@googlegroups.com <mailto:raxml+un...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/raxml/e9bbffb7-1c84-4b0c-b8ac-634bc74a8d2b%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/e9bbffb7-1c84-4b0c-b8ac-634bc74a8d2b%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Miao Sun]

Hi Alexey,

Thanks for your response!

You are right. The alignment is generated from upstream pipeline. It's just one of hundred gene alignments, which occasionally contains a few samples. I will put this gene tree (if succeed) along with other gene trees to feed ASTRL (my plan). 

Do you think it there way somehow I can finish tree building for this gene to continue my plan? Or are you suggesting I need to exclude this gene because of poor phylogenetic signal?

Thanks!

Miao

Dr. Miao Sun

Web: www.sunmiao.name                 Email: mia...@bios.au.dk

You received this message because you are subscribed to a topic in the Google Groups "raxml" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/raxml/INJNA4SlbDI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to raxml+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/raxml/0cffdc26-4f45-6f58-1903-b97429cede77%40gmail.com.

[Alexey Kozlov]

yes you should definitely exclude gene trees with that few sites. I'd say anything below 50 sites is
pretty useless, although it's better to check the literature on this topic (ML gene trees + ASTRAL
is a widely-used pipeline).

maybe you should also check your alignment filtering setting, since I'm pretty sure the original
gene MSA had more than 3 sites...

On 10.12.19 14:15, Miao Sun wrote:
> Hi Alexey,
> Thanks for your response!
>
> You are right. The alignment is generated from upstream pipeline. It's just one of hundred gene
> alignments, which occasionally contains a few samples. I will put this gene tree (if succeed) along
> with other gene trees to feed ASTRL (my plan).
>
> Do you think it there way somehow I can finish tree building for this gene to continue my plan? Or
> are you suggesting I need to exclude this gene because of poor phylogenetic signal?
>
> Thanks!
>
> Miao
>
>
> Dr. Miao Sun

> Web: www.sunmiao.name <http://www.sunmiao.name>                 Email: mia...@bios.au.dk
> <mailto:mia...@bios.au.dk>

> > raxml+un...@googlegroups.com <mailto:raxml%2Bunsu...@googlegroups.com>
> <mailto:raxml+un...@googlegroups.com <mailto:raxml%2Bunsu...@googlegroups.com>>.

> > To view this discussion on the web visit
> > https://groups.google.com/d/msgid/raxml/e9bbffb7-1c84-4b0c-b8ac-634bc74a8d2b%40googlegroups.com
> >
> <https://groups.google.com/d/msgid/raxml/e9bbffb7-1c84-4b0c-b8ac-634bc74a8d2b%40googlegroups.com?utm_medium=email&utm_source=footer>.
>
> --
> You received this message because you are subscribed to a topic in the Google Groups "raxml" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/raxml/INJNA4SlbDI/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to

> raxml+un...@googlegroups.com <mailto:raxml%2Bunsu...@googlegroups.com>.

> To view this discussion on the web visit

> https://groups.google.com/d/msgid/raxml/0cffdc26-4f45-6f58-1903-b97429cede77%40gmail.com.

>
> --
> You received this message because you are subscribed to the Google Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to
> raxml+un...@googlegroups.com <mailto:raxml+un...@googlegroups.com>.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/raxml/CAM9zNnj3MeEt6UEQA3PjMqqSSBZNqJiA6xTyRu3SqY65dEyS-Q%40mail.gmail.com
> <https://groups.google.com/d/msgid/raxml/CAM9zNnj3MeEt6UEQA3PjMqqSSBZNqJiA6xTyRu3SqY65dEyS-Q%40mail.gmail.com?utm_medium=email&utm_source=footer>.

[Miao Sun]

Thanks!!! You comments are very helpful!!

Miao

Dr. Miao Sun

Web: www.sunmiao.name                 Email: mia...@bios.au.dk

To unsubscribe from this group and all its topics, send an email to raxml+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/raxml/2f04c4b0-6bce-5424-d773-016aa91e65a9%40gmail.com.