ASC_GTRCAT finds invariant sites that are not in the input

[Emiliano Trucchi]

Hi all,

Using the option to correct for invariant sites via ascertainment bias (-m ASC_GTRCAT) in raxmlHPC-AVX, I got an error saying there are invariant sites left.

"For partition No Name Provided you specified that the likelihood score shall be corrected for invariant sites
via an ascertainment bias correction. However, some sites in this partition are already invariant.
This is not allowed, please remove all invariant sites and try again, exiting ... "

Any idea about it?

Looking at the very small test alignment attached, it doesn't seem there are invariants. I read on this forum that ambiguous bases are NOT considered as missing data, is that correct?

Thanks a lot.

All the best,
Emiliano

[Alexandros Stamatakis]

Hi Emiliano,

Which version are you using?

In general if you have a column with A's but also a couple of ambiguous
characters including A, then this will also be considered as an
invariant site, I believe that I had added some more verbose error
output in version 8.2.4,

Alexis

> --
> 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>.
> For more options, visit https://groups.google.com/d/optout.

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
Adjunct Professor, Dept. of Ecology and Evolutionary Biology, University
of Arizona at Tucson

www.exelixis-lab.org

[Emiliano Trucchi]

Hi Alexis,

I was using version  8.1.3 . I will update and let you know.
Does it mean that ambiguous characters are treated as missing data? I thought they were assigned an equal probability of being one or the other base.
Actually I have a lot of them so if they are not taken into account my matrix will be way less informative.

Thanks for the help

Best,
Emiliano

[Alexandros Stamatakis]

dear emiliano,

yes please upgrade to v 8.2.4

> Does it mean that ambiguous characters are treated as missing data?

yes, in most cases

> I
> thought they were assigned an equal probability of being one or the
> other base.

that's correct, they are actually all assigned a probability of 1 ...

> Actually I have a lot of them so if they are not taken into account my
> matrix will be way less informative.

well yes, but the maths break down when you have such invariant sites
with ambiguous characters, hence they have to be removed ...

alexis

> > an email to raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml+un...@googlegroups.com <javascript:>>.

> > For more options, visit https://groups.google.com/d/optout

> <https://groups.google.com/d/optout>.

>
> --
> Alexandros (Alexis) Stamatakis
>
> Research Group Leader, Heidelberg Institute for Theoretical Studies
> Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
> Adjunct Professor, Dept. of Ecology and Evolutionary Biology,
> University
> of Arizona at Tucson
>

> www.exelixis-lab.org <http://www.exelixis-lab.org>

[Emiliano Trucchi]

Hi Alexis,

Thanks a lot for the clarification.

Best,
Emiliano

[Amber Bass]

Hi Alexis,

What is the best way to quickly remove these invariant sites in a data set with thousands of SNPs?

Thanks,
Amber

[Alexandros Stamatakis]

Hi Amber,

Somebody had posted a script on the google group 3-4 years ago, I just
recently searched for this again and re-posted it. You should be able to
find it in the posts of the last 2 months where we had a similar discussion.

Alexis

www.exelixis-lab.org