You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
|`--link-exchange-rates`| Turn on linked exchangeability estimation for a profile mixture model. Note that the model must have specified `GTR20` exchangeabilities for eg.`GTR20+C20+G`. |
332
+
|`--gtr20-model`| Specify the initial exchangeabilities for linked exchangeability estimation. Note that this must be used with `--link-exchange-rates.`|
333
+
|`--rates-file`| Produces a nexus file with the exchangeability matrix obtained from the optimization. This file can be later used for phylogenetic inference with the use of the `-mdef` flag |
334
+
335
+
### Example usages:
336
+
337
+
* Estimate linked exchangeabilities for a protein alignment `prot.phy` under C60+G model and a guide tree `guide.treefile`, where optimization is initialized from LG exchangeabilities
338
+
339
+
iqtree -s prot.phy -m GTR20+C60+G --link-exchange-rates --gtr20-model LG -te guide.treefile
340
+
341
+
>**NOTE**: For better and faster performance, read the [recommendations](Complex-Models#linked-gtr-exchangeabilities-models) provided in the Complex Models section.
In this example exchangeabilities will be estimated for a profile mixture model `C60+G4` but any profile mixture model and rates can be used. To estimate a single set of linked exchangeabilities, in the model definition the matrix `GTR20` must be specified (resp. GTR for nucleotide data) together with the flag `--link-exchange-rates`. While a guide tree is not needed, we highly recommend using a fixed tree topology to estimate exchangeabilities. Since matrix estimation can be time-consuming, we also recommend using the flag `-me 0.99` to reduce the optimization threshold for faster optimization. Simulations have shown that changing this parameter has no significant effect on exchangeability estimation.
213
+
214
+
The user can determine the starting exchangeabilities before optimization. Choosing adequate exchangeabilities can make estimation considerably faster. For example:
specifies the LG matrix as the starting matrix via the flag `--gtr20-model` (the default starting matrix is POISSON, i.e. equal exchangeabilities). For this flag, the user can specify any matrix, even those matrices defined by the user via the `-mdef` flag. If the user is agnostic of the exchangeabilities, we recommend using the default matrix (although it can be time-consuming).
219
+
220
+
Note that the user can estimate exchangeabilities jointly with weights of the profiles, branch lengths, and rates. This can be very time-consuming. If the goal is to optimize exchange abilities, one can fix the other parameters to reasonable estimates (for eg. fixing branch lengths and rates has been shown to perform adequately for estimation of exchangeabilities)
221
+
222
+
There is an additional flag `--rates-file` that will produce a nexus file with the exchangeability matrix obtained from the optimization. This file can be later used for phylogenetic inference with the use of the `-mdef` flag.
223
+
224
+
225
+
If you use this routine in a publication please cite:
226
+
227
+
> __H. Banos et al.__ (2024) Estimating Linked Exchangeabilities for Profile Mixture Models. _Bioraxiv.
228
+
229
+
198
230
Here, the NEXUS file contains a `models` block to define new models. More explicitly, we define four AA profiles `Fclass1` to `Fclass4`, each containing 20 AA frequencies. Then, the frequency mixture is defined with
0 commit comments