search

CompEvol / beast2

Bayesian Evolutionary Analysis by Sampling Trees
www.beast2.org
GNU Lesser General Public License v2.1
236 stars 83 forks source link

Port sequencing error models from BEAST1 to BEAST2 #173

Open alexeid opened 10 years ago

alexeid commented 10 years ago

Sequencing error models (and post-mortem damage models) are implemented in BEAST1. See http://mbe.oxfordjournals.org/content/26/2/245.full Users would like this ported to BEAST2.

ddkapan commented 2 years ago

Any update on generalized DNA damage handling a la http://mbe.oxfordjournals.org/content/26/2/245.full?

kche309 commented 2 years ago

@ddkapan I have been working on an error model framework for beast2 which includes a nucleotide sequencing error model (in testing phase). It is slightly different from the Oxford paper in that we did not use an exponential.

www.github.com/bioDS/beast-phylonco

I can add the exact model from the Oxford paper. This should take about a day, thorough testing may take 1-2 weeks.

Please wait for the new release v0.0.5 which will include new models, faster implementation and bug fixes.

I'll update you as things progress :)

ddkapan commented 2 years ago

Hi,

That’s great news.

I’m interested in both sources of error, but using NGS on >75 year old museum specimens, post-mortem DNA damage modeled as a function of time is more likely to bias Ts vs Tv (a pairwise distance matrix built from transversions shrinks the branch lengths, likely more so for older specimens, so it may even work on such short time frames).

I’m excited to use the exponential PMD model in beast-phylonco when it is available. I’m fairly new to BEAST2 but I’ll happily run it and share my findings if it helps. Thanks!

All the best,

Durrell

On Oct 21, 2021, at 4:36 PM, kche309 @.***> wrote:

 @ddkapan I have been working on an error model framework for beast2 which includes a nucleotide sequencing error model (in testing phase). It is slightly different from the Oxford paper in that we did not use an exponential.

www.github.com/bioDS/beast-phylonco

I can add the exact model from the Oxford paper. This should take about a day, thorough testing may take 1-2 weeks.

Please wait for the new release v0.0.5 which will include new models, faster implementation and bug fixes.

I'll update you as things progress :)

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

ddkapan commented 2 years ago

Hi, @kche309 any update on adding the PMD damage model and in general when the new release v0.0.5 will come out? Although I'm not sure I have the proficiency to help code the model, I'd love to help with the test of the update. Thanks! Durrell

kche309 commented 2 years ago

Hi @ddkapan, We're currently working on getting the user interface working (Beauti). There's currently a bug in the templates which means configuring these types of analyses will require manual editing of xml files (https://github.com/bioDS/beast-phylonco/issues/8).

We will be adding the PMD model as soon as the UI issue is resolved.

ddkapan commented 2 years ago

Hi @kche309 that's awesome. Once you have the UI issue resolved and time to add in the PMD model, I'll be happy to try out the code and report back. I really appreciate your efforts!

kche309 commented 2 years ago

Hi @ddkapan just a brief update on expected timeframe for the PMD model. We will schedule to implement this in April. I will let you know once an early version is available.

ddkapan commented 2 years ago

Hi kche309, Thanks! That sounds perfect as we should have all our data in hand by then. Durrell  On Tuesday, March 1, 2022, 11:54:34 AM PST, kche309 @.***> wrote:

Hi @ddkapan just a brief update on expected timeframe for the PMD model. We will schedule to implement this in April. I will let you know once an early version is available.

— Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you were mentioned.Message ID: @.***>

ddkapan commented 2 years ago

Hi sorry to be a pest, any further updates would be appreciated. Happy to test it out as well! :) Durrell

kche309 commented 2 years ago

@ddkapan Hi work is in progress for this model on branch https://github.com/bioDS/beast-phylonco/tree/PMDErrorModel The main source code is in PMDErrorModel.java This is pending a fix on the Gradle build file testing and a few TODOs.

@walterxie thanks for your help on the Gradle fix.

kche309 commented 2 years ago

@ddkapan I have a few quick questions about your data. (1) Do you have a mix of extant species and museum fossil samples? (2) Would it be correct to assume you have a list of taxa names for your fossil data (potentially affected by post-mortem damage)?

kche309 commented 2 years ago

@ddkapan A brief update, I've added an exclusion option (for excluding certain taxa from the PMD error model, e.g. extant species).

Main code is now complete, I will be adding some unit tests and working on the XML template 😄

ddkapan commented 1 year ago

HI @kche309, I'm sorry I missed your note on my GitHub notifications. Awesome work (!) and yes we will have taxa from museum samples with PMD and modern samples where it is not expected so that's perfect.. I also see the notes atbeast-phylonco. Now I'll need to figure out how to use the tools. Thank you!