This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Best practices and parameter optimization for aligning highly degraded ancient S. mutans genomes?

Hello everyone,

I am currently working on variant calling and phylogenetic analysis of ancient Streptococcus mutans genomes using an HPC environment (SLURM).

Initially, we attempted to process the datasets using standard modern NGS pipelines, which led us in the wrong direction due to the inherent characteristics of aDNA (short fragment lengths and deamination patterns). We are now restructuring our methodology specifically for ancient metagenomic data.

I have a few methodological questions:

Which aligner (and specific parameters) do you strongly recommend for bacterial aDNA to properly handle short reads without losing true variants?

What are the most robust tools/pipelines you use to validate damage patterns (e.g., mapDamage) before moving on to variant calling?

Any advice on transitioning a pipeline from modern DNA to aDNA standards would be highly appreciated.

For more context on our current approach and the specific scripts we tested on our HPC, you can check our project repository here: https://github.com/oguzhanisilay8/S_mutans_Variant_and_Phylogeny_Analysis

bwa alignment variant-calling metagenomics ancient-dna
• 218 views • link
written by oguzhanisilay • oguzhanisilay

In what way are these "ancient"? Are they preserved samples or from some special environment? Are the libraries being prepared in any special way?

to properly handle short reads

What is the length of the reads?

• 0 views • link
written by GenoMax GenoMax

Hi, thanks for the clarifying questions. To provide more context on the dataset and our workflow:

  1. Origin and Damage Profiles: The samples are strictly archaeological. Based on our metadata, their mean age estimates range broadly from ~200 to 7,800 years BP (covering Mesolithic to Historical contexts across Eurasia). Our metadata confirms characteristic aDNA damage, with metaDMG Bayesian Z scores ranging from roughly 2.9 to over 11.5.

  2. Library Prep and Read Lengths: The core of this project was to retrieve raw datasets from various original publications, build a unified, reproducible bioinformatics pipeline, and execute it on our national HPC cluster (TRUBA). Because these raw datasets were compiled from different original studies (the foundational dataset references, including Thygesen et al., 2026, can be found in our repository), the exact library preparation protocols (e.g., UDG treatments) and initial read lengths vary across the samples.

However, after strict quality control and adapter trimming on our end, a substantial portion of the reads consistently falls into the short fragment range typical for highly degraded aDNA (well below 50 bp).

Given this heterogeneity in the raw data sources, the extreme degradation, and the specific damage profiles we need to retain, would you recommend sticking to bwa aln (for instance, with disabled seed length -l 16500) to maximize the mapping of these short, damaged fragments across such a diverse dataset?

• 0 views • link
written by oguzhanisilay • oguzhanisilay

Sounds like you already have the pipeline locked down well. You may have already seen this pipeline meant for analysis of aDNA: https://nf-co.re/eager/2.5.3 but if not you can take a look.

• 0 views • link
written by GenoMax GenoMax

Thank you for the recommendation and the encouraging feedback!

• 0 views • link
written by oguzhanisilay • oguzhanisilay

0 answers

No answers yet.

Log in to answer this question.

More posts like this