Multiple character states at same locus

21 views
Skip to first unread message

S

unread,
Aug 24, 2016, 4:49:29 PM8/24/16
to raxml
Is there a way to model multi-state characters where there are several characters present at the same locus/ slot ?
For example, if my data set is for morphological data, the Z corresponds to both character states 1 and 2 being simultaneously present and the X correspond to states 1, 2 and being present. Is there a way to model this in Raxml?
t1	0110Z11100 0001000?1? 0000000000 000010Z110 0110102324
t2	0110111100 0001001110 0000000000 0000100100 1010101234
t3	0110111100 0001001110 0000000000 0000100100 1010102013
t4	0110111100 0001001110 000X000000 0000100100 1010103102
t5 	0110111110 0010110110 111?0?0000 0000110000 0101002102
t6 	0110111110 0010110110 1111001000 0000010000 0100013211
t7 	0110111110 0010110101 1111000000 0010100000 0100001233
t8	0110111110 0010110111 111?001000 0010100000 0100102311
t9	0110111110 0010110101 11010??000 0000100000 0100102211
t10 	0110111110 0010110010 0001011000 0000100000 0100101144

Alexey Kozlov

unread,
Aug 24, 2016, 6:17:44 PM8/24/16
to ra...@googlegroups.com
Hi Steven,

unfortunately, Raxml doesn't support ambiguous characters for multistate data.

The only workaround I can suggest is to "unroll" your states with binary characters, e.g.

0 = 0000
1 = 0001
2 = 0010
3 = 0100
4 = 1000
Z = 0011
X = 0111
? = 1111

(or use 5 bits/slot if you consider 0 as a separate state).

Then you'd get a binary alignment and so can use raxml with BINGAMMA model.

Hope this helps,
Alexey

On 24.08.2016 22:49, S wrote:
> Is there a way to model multi-state characters where there are several characters present at the same locus/ slot ?
> For example, if my data set is for morphological data, the Z corresponds to both character states 1 and 2 being
> simultaneously present and the X correspond to states 1, 2 and being present. Is there a way to model this in Raxml?
>
> t1 0110*Z*11100 0001000?1? 0000000000 000010*Z*110 0110102324
> t2 0110111100 0001001110 0000000000 0000100100 1010101234
> t3 0110111100 0001001110 0000000000 0000100100 1010102013
> t4 0110111100 0001001110 000*X*000000 0000100100 1010103102
> t5 0110111110 0010110110 111?0?0000 0000110000 0101002102
> t6 0110111110 0010110110 1111001000 0000010000 0100013211
> t7 0110111110 0010110101 1111000000 0010100000 0100001233
> t8 0110111110 0010110111 111?001000 0010100000 0100102311
> t9 0110111110 0010110101 11010??000 0000100000 0100102211
> t10 0110111110 0010110010 0001011000 0000100000 0100101144
>
>
> --
> 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.
Reply all
Reply to author
Forward
0 new messages