Skip to content

Commit cbb9360

Browse files
bqminhbqminh
committed
update molevol2025 by Hanon
1 parent 292b26d commit cbb9360

1 file changed

Lines changed: 55 additions & 46 deletions

File tree

‎workshop/molevol_tutorial2025.md‎

Lines changed: 55 additions & 46 deletions
Original file line numberDiff line numberDiff line change
@@ -29,7 +29,7 @@ IQ-TREE 3 Tutorial (Workshop on Molecular Evolution, Woods Hole 2025)
2929

3030
<!--more-->
3131

32-
In the virtual machine established by the organisers you can run IQ-TREE version 2.2.2.6
32+
In the virtual machine established by the organizers you can run IQ-TREE version 3.0.1
3333
from the command line:
3434

3535
iqtree3
@@ -254,49 +254,6 @@ Options explained:
254254
{: .tip}
255255

256256

257-
X) Applying a mixture model
258-
---------------------------
259-
<div class="hline"></div>
260-
261-
262-
We now perform a mixture model analysis ([Ren et al., 2025]), where one allows
263-
each site to be described as a mixture of models.
264-
265-
> What is the command line to run `iqtree3` that takes `turtle.fa` as input alignment,
266-
> estimates the phylogeny under the best model from step 1, but now as mixture model with two separate classes?
267-
> This command should also perform 1000 ultrafast bootstrap replicates and automatically determine the best number of cores.
268-
269-
<button type="button" class="btn btn-primary" id="buttonX" onclick="myFunction('buttonX','commandX')">
270-
Show me the command line</button>
271-
<div id="commandX" style="display: none; border:1px solid gray;">
272-
273-
<pre><code>iqtree3 -s turtle.fa -m MIX"{GTR+FO,GTR+FO}"+I+R3 -B 1000 --prefix turtle2</code></pre>
274-
275-
Options explained:
276-
277-
<ul>
278-
<li>`-m MIX"{GTR+FO,GTR+FO}"+I+R3` to specify a mixture model with two separate GTR models ([Ren et al., 2025]) with optimized nucleotide frequencies as well as invariant sites and 3 freerate categories. The quotes aroung the curly braces are important, otherwise bash will try to do variable expansion.</li>
279-
<li>`--prefix turtle2` to keep file names informative and tidy. </li>
280-
</ul>
281-
282-
> Now use the same command structure to try mixture models with 3, 4, and 5 classes!
283-
284-
</div>
285-
286-
287-
> **QUESTIONS:**
288-
>
289-
> * Look at the report file `turtle2.iqtree`, 'turtle3.iqtree', etc. What are the AIC/AICc/BIC scores of the various mixture models? Is it better than a 1-class model?
290-
>
291-
> * Look at the tree in `turtle2.iqtree` or visualize `turtle2.treefile`
292-
> in FigTree. What relationship among [three trees](#1-input-data) does this tree support? What about for the 3-, 4-, and 5-class models?
293-
>
294-
> * What is the ultrafast bootstrap support (%) for the relevant clade under each model?
295-
>
296-
> * Which of these trees agree with the published tree ([Chiari et al., 2012])?
297-
{: .tip}
298-
299-
300257
5) Tree topology tests
301258
----------
302259
<div class="hline"></div>
@@ -411,7 +368,7 @@ Options explained:
411368

412369
> **QUESTIONS:**
413370
>
414-
> * Look at `turle.mix.iqtree` for the line printing the tree weights. Which tree has a higher weight?
371+
> * Look at `turtle.mix.iqtree` for the line printing the tree weights. Which tree has a higher weight?
415372
> * Is it the tree having higher likelihood found in step 5?
416373
{: .tip}
417374

@@ -461,7 +418,7 @@ log-likelikehood subtraction as pointed out above).
461418
We now try to construct a tree without these "influential" genes.
462419
To do so, copy the partition file `turtle.nex` to a new file and
463420
remove the lines defining the `charset` of these genes, and then
464-
repeat the IQ-TREE run with a parititon model (see section 4).
421+
repeat the IQ-TREE run with a partition model (see section 4).
465422
You will need to figure out a command line to run IQ-TREE yourself here.
466423

467424
> **QUESTIONS:**
@@ -546,11 +503,62 @@ Similarly, you can compute gCF and sCF for the tree under unpartitioned model:
546503
> contradicting branch?
547504
{: .tip}
548505

506+
507+
10) Applying a mixture model
508+
---------------------------
509+
<div class="hline"></div>
510+
511+
512+
We now perform a mixture model analysis ([Ren et al., 2025]), where
513+
each site is described as a mixture of models. For this exercise we will use a slightly different model, GTR+FO+I+R3, which is the best fit to the entire turtle dataset.
514+
515+
Estimate a phylogeny under this model as a baseline:
516+
517+
iqtree3 -s turtle.fa -m GTR+FO+I+R3 -B 1000 -T AUTO --prefix turtlebest
518+
519+
> What is the command line to run `iqtree3` that takes `turtle.fa` as input alignment and
520+
> estimates the phylogeny under GTR+FO+I+R3, but now as mixture model with two separate classes?
521+
> This command should also perform 1000 ultrafast bootstrap replicates and automatically determine the best number of cores.
522+
523+
<button type="button" class="btn btn-primary" id="buttonX" onclick="myFunction('buttonX','commandX')">
524+
Show me the command line</button>
525+
<div id="commandX" style="display: none; border:1px solid gray;">
526+
527+
<pre><code>iqtree3 -s turtle.fa -m MIX"{GTR+FO,GTR+FO}"+I+R3 -B 1000 -T AUTO --prefix turtle2</code></pre>
528+
529+
Options explained:
530+
531+
<ul>
532+
<li>`-m MIX"{GTR+FO,GTR+FO}"+I+R3` to specify a mixture model with two separate GTR models ([Ren et al., 2025]) with optimized nucleotide frequencies as well as invariant sites and 3 freerate categories. The quotes aroung the curly braces are important, otherwise bash will try to do variable expansion. </li>
533+
<li>`--prefix turtle2` to keep file names informative and tidy. </li>
534+
</ul>
535+
536+
> Now use the same command structure to try a mixture model with 4 classes, and another one with 6 classes! These will take some time to run, more as the number of classes increases. In the meantime, check out [Ren et al., 2025], which performed a similar analysis of mixture models with different numbers of model classes - on a larger turtles dataset.
537+
538+
</div>
539+
540+
541+
> **QUESTIONS:**
542+
>
543+
> * Look at the report files `turtlebest.iqtree`, `turtle2.iqtree`, `turtle4.iqtree`, and `turtle6.iqtree`. What are the BIC scores of the various mixture models? Which is best?
544+
>
545+
> * Look at the tree in `turtle2.iqtree` or visualize `turtle2.treefile`
546+
> in FigTree. What relationship among [three trees](#1-input-data) does this tree support? What about for the 4- and 6-class models?
547+
>
548+
> * What is the ultrafast bootstrap support (%) for the relevant clade under each model?
549+
>
550+
> * Which of these trees agree with the published tree ([Chiari et al., 2012])? Do the relationships or ultrafast bootstrap values change with model complexity?
551+
> * [Ren et al., 2025] performed a similar analysis of the effect of number of model classes on the relationships between turtle, bird, and crocodile. What relationship among [three trees](#1-input-data) do these authors find supported by a 1-class model? What about a 6-class model? Do our results agree?
552+
{: .tip}
553+
554+
549555
> **FINAL QUESTIONS:**
550556
>
551557
> * Given all analyses you have done in this tutorial, which relationship between
552558
> Turtle, Crocodile and Bird is true in your opinion?
553559
560+
561+
554562
[Adachi and Hasegawa, 1996]: http://www.is.titech.ac.jp/~shimo/class/doc/csm96.pdf
555563
[Anisimova et al., 2011]: https://doi.org/10.1093/sysbio/syr041
556564
[Brinkmann et al., 2005]: https://doi.org/10.1080/10635150500234609
@@ -573,6 +581,7 @@ Similarly, you can compute gCF and sCF for the tree under unpartitioned model:
573581
[Minh et al., 2020]: https://doi.org/10.1093/molbev/msaa106
574582
[Nei et al., 2001]: https://doi.org/10.1073/pnas.051611498
575583
[Nguyen et al., 2015]: https://doi.org/10.1093/molbev/msu300
584+
[Ren et al., 2025]: https://doi.org/10.1093/molbev/msae264
576585
[Shimodaira and Hasegawa, 1999]: https://doi.org/10.1093/oxfordjournals.molbev.a026201
577586
[Shimodaira, 2002]: https://doi.org/10.1080/10635150290069913
578587
[Strimmer and Rambaut, 2002]: https://doi.org/10.1098/rspb.2001.1862

0 commit comments

Comments
 (0)