Comments (11)
Dear @glitterseaworms,
Trust neither. There is a serious issue with the alignment itself: probably lack of homology. You can see this because many branch lengths in the file are infinite (the total tree length is > 10,000 subs/site!).
![image](https://private-user-images.githubusercontent.com/1018513/348788743-6234ad2a-9ca8-4600-be49-5f4dd76d33aa.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.sAYjclYQGkZywA90Po8pmN1JLUGksg3KRDccK3VKIBM)
With such alignments, where sequences look so diverged as to be seemingly unrelated, all inference becomes very unstable and unreliable.
Would you share your input alignment here so that I could take a closer look?
Best,
Sergei
from hyphy.
Dang, my data is really giving me a hard time. This is such a bummer. I think my worms in general are just very divergent. What is a tree length subs/site value cut off that you would suggest is reasonable to trust results at?
I am having trouble attaching the alignments here, as it says file type not supported--can I please email them to you?
Thank you for your dedication and time.
from hyphy.
Hi Sergei, I think I may have solved the issue. It looks like if I reduce my tree and alignments to the family taxon level as opposed to representatives from the whole order that I had included, this alleviates the subs/site issue. For ATP6 gene for instance, I am now receiving a subs/site value of 9.956 as opposed to 7207.802 with the larger dataset. I will keep you posted as tests finish, but I think I will proceed by re-running everything with the reduced trees.
from hyphy.
Dear @glitterseaworms,
Thanks for the update! Could you attach the reduced ATP6
(as a .txt
) file? I'll take a quick look to make sure nothing looks suspicious.
I took a look at the full (56 sequences) ATP6
file, and if you use raxml-ng
to infer the ML tree, its total length is ~12 (~21 with codon models), which is not unreasonable. What have you been using to build the trees?
Here's what BUSTED-E
on it looks like. I typically use it as a screen to look for oddities...
hyphy busted --alignment ATP6_codon_aligned.fas --tree ATP6_codon_aligned.fas.raxml.bestTree --code Invertebrate-mtDNA --intermediate-fits ATP6.fit --error-sink Yes --output ATP6.BUSTED-E.json
....
* Log(L) = -22981.63, AIC-c = 46236.08 (135 estimated parameters)
* For *test* branches, the following rate distribution for branch-site combinations was inferred
| Selection mode | dN/dS |Proportion, %| Notes |
|-----------------------------------|---------------|-------------|-----------------------------------|
| Negative selection | 0.001 | 67.171 | |
| Negative selection | 0.018 | 32.479 | |
| Neutral evolution | 1.000 | 0.257 | |
| Error absorption | 100.000 | 0.094 | |
* The following rate distribution for site-to-site **synonymous** rate variation was inferred
| Rate | Proportion, % | Notes |
|-----------------------------------|---------------|-----------------------------------|
| 0.079 | 53.394 | |
| 0.725 | 24.993 | |
| 3.593 | 21.613 | |
### No evidence for episodic diversifying positive selection under the unconstrained model, skipping constrained model fitting
----
## Branch-site unrestricted statistical test of episodic diversification [BUSTED]
Likelihood ratio test for episodic diversifying positive selection, **p = 0.5000**.
Use https://observablehq.com/@spond/busted to visualize the resulting JSON
![image](https://private-user-images.githubusercontent.com/1018513/350791944-a47db0aa-9968-4c6c-9c1d-33b00582e5c4.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.lKX19qRuBikCroVN1m5C3UzNu2Oc_YGNG98igaQWtbw)
![image](https://private-user-images.githubusercontent.com/1018513/350792029-3d2d1d79-6755-4e7c-b23b-b21e497d7038.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.XxNxx_w6O9meRcTvHAnPqALO1Riuet58fcDZaZfsviA)
![image](https://private-user-images.githubusercontent.com/1018513/350792101-a3734f36-c788-4c21-ab9a-9ee62f112d41.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.Nm13F3gbGK434v12Y_xv3ash9TFq8hcEcw4yspk4Efg)
Looks like you have a couple of species that are highly diverged, possible sequencing/assembly errors because there are things like 3x substitutions along a single branch (like the red branch in the tree), and runs of sequences where there are may be some homology issues (or they are just too diverged).
Best,
Sergei
from hyphy.
Hi Sergei,
Thanks so much for responding very thoroughly. I truly appreciate you and will be acknowledging you in my manuscript :-) This means a lot and has helped me better understand results and how to properly work with my dataset.
I use RAxML to build to input trees. Dang, not sure what is going on with the red branch, but that species sequence should hypothetically be correct. I do think my clade of interest is very divergent in general. I have attached the reduced ATP6 alignment. Please let me know if anything seems strange with it. All genes seem to run properly with the reduced dataset, so I am going to re-run all 13 genes with aBSREL, Busted and Meme using this reduced dataset instead.
Looking forward to hearing your thoughts on my alignment, and truly, thanks again.
Best,
Avery
ATP6_reduced_codon_aligned.txt
from hyphy.
Hello again,
I just wanted to follow up regarding the issue of different results when running replicates. I just finished running 4 replicates of aBSREL with my reduced data set. All subs/site values are values that can be trusted. However, for 6 of the 13 genes I am testing, I obtain variation in results across the replicates. I've attached the output files for these genes. If you have the ability, can you please see if this is normal? If so, my plan was to only report results that are the same across the 4 replicates, and dismiss other branches that show up.
Thanks again,
Avery
ATP6_aBSREL_Deep_rep0_output.txt
ATP6_aBSREL_Deep_rep1_output.txt
ATP6_aBSREL_Deep_rep2_output.txt
ATP6_aBSREL_Deep_rep3_output.txt
COX2_aBSREL_Deep_rep0_output.txt
COX2_aBSREL_Deep_rep1_output.txt
COX2_aBSREL_Deep_rep2_output.txt
COX2_aBSREL_Deep_rep3_output.txt
ND3_aBSREL_Deep_rep0_output.txt
ND3_aBSREL_Deep_rep1_output.txt
ND3_aBSREL_Deep_rep2_output.txt
ND3_aBSREL_Deep_rep3_output.txt
ND4_aBSREL_Deep_rep0_output.txt
ND4_aBSREL_Deep_rep1_output.txt
ND4_aBSREL_Deep_rep2_output.txt
ND4_aBSREL_Deep_rep3_output.txt
ND5_aBSREL_Deep_rep0_output.txt
ND5_aBSREL_Deep_rep1_output.txt
ND5_aBSREL_Deep_rep2_output.txt
ND5_aBSREL_Deep_rep3_output.txt
ND6_aBSREL_Deep_rep0_output.txt
ND6_aBSREL_Deep_rep1_output.txt
ND6_aBSREL_Deep_rep2_output.txt
ND6_aBSREL_Deep_rep3_output.txt
from hyphy.
Dear @glitterseaworms,
Could you attach the alignment & tree for one of the datasets where you are seeing discordant results (e.g. ATP6)?
Best,
Sergei
from hyphy.
Attached below:
4NT_14AA_Tree_48terminals_allloci_500mDeep.txt
ATP6_reduced_codon_aligned.txt
from hyphy.
Hi Sergei,
Not sure if that's affecting it, but my tree was a mess. I am going to try using the attached tree instead with no branch lengths and unrooted. I'll run 4 replicates and see if I have the same issue. I'll keep you posted.
Avery
4NT_14AA_Tree_48terminals_allloci_unrooted_noBL_deep.txt
from hyphy.
Dear @glitterseaworms,
Ah, it looks like the default optimization precision is insufficient for this dataset (probably still too divergent). If you see negative LRTs (especially <-1) in the testing table, this is an indication. Let me look into why this happening. A few days.
Best,
Sergei
from hyphy.
Thanks Sergei! In case this is useful for the investigation in some way, I tried running ATP6 with an unrooted input tree containing branch lengths and one without branch lengths previously specified. Both outputs are attached. As you can see, aBSREL analysis with branch lengths on input tree results in subs/site 9.955, but without branch lengths on input tree results in 1583.030. However, both outputs contain -LRTs in the testing tables.
Attached are the 2 different trees I used, and the 2 output files. Hopefully this helps. Thank you again!!
4NT_14AA_Tree_48terminals_allloci_unrooted_noBL_deep.txt
4NT_14AA_Tree_48terminals_allloci_unrooted_yesBL_deep.txt
ATP6_aBSREL_Deep_unrooted_noBL_rep0_output.txt
ATP6_aBSREL_Deep_unrooted_yesBL_rep0_output.txt
from hyphy.
Related Issues (20)
- What is HyPhy calling the 2 nodes pointed at with arrows in this tree (image and txt file attached)? HOT 2
- aBSREL run time HOT 5
- Persistent error when running Gard HOT 5
- How to select positive signals of the foreground branch from the results of aBSREL modle HOT 4
- error HOT 7
- Analysis ends but does not progress to report generation HOT 5
- Failed to dereference error with FitMG94 and the local option HOT 2
- test for detecting compartmentalization in dataMonkey HOT 2
- DataMonkey not recognizing forground branches HOT 2
- MEME interpretation - 0 branches? HOT 4
- How to process the results files from OrthoFinder into input files for HyPhy ABSREL? HOT 9
- How to choose background branches? HOT 2
- Does excluding species with paralogous sequences violate aBSREL and/or RELAX assumptions of single-copy orthology? HOT 6
- Interpretation of RELAX across many genes HOT 2
- FADE: substitution model selection
- Multi-ORF Analysis Questions
- Barely any parallelism (max 110% CPU) running `hyphy gard` on macOS arm64 HOT 1
- GARD crashes when run on 27 SARS-CoV-2 sequences HOT 4
- Format for selecting branches to test on FEL HOT 8
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from hyphy.