This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Resources for understanding variant calling pipeline outputs

Hi everyone,

I am working on a variant calling project for learning purposes and I am struggling to understand and interpret the output from each step. I want to know what each file and result means not just how to generate it.

To give a simple example, after mapping reads to the reference genome you usually run samtools flagstat. I would like to understand which lines in the output are the most important to pay attention to, what kinds of sanity checks I should be doing and what it means if a particular metric is higher or lower than normal.

I have already searched through BioStars but I could not find answers to many of the questions I have.

If you know of any books, blogs or paper that explain how to interpret the output at each step, I would really appreciate the recommendations. Thanks!

variant calling ngs
• 141 views • link
updated by Prash • Prash • written by ptalebic • ptalebic

2 answers

I could not find answers to many of the questions I have.

If you have specific questions then feel free to ask them directly Simple searches on biostars, sometimes may not identify the right thread. You can also try asking your favorite AI/LLM.

GATK has a collection of posts about best practices for variant discovery here: https://gatk.broadinstitute.org/hc/en-us/sections/360007226651-Best-Practices-Workflows

Harvard Chan Bioinformatics core has an Intro course available here: https://github.com/hbctraining/Intro-to-variant-analysis/blob/main/schedule/self-learning.md

While the galaxy courses are meant for use on galaxy they should have the basics you are asking about covered: https://training.galaxyproject.org/training-material/topics/variant-analysis/

• 0 views • link
written by GenoMax GenoMax

GenoMax Thanks for the reply and for sharing those great resources.

After aligning my reads, I started looking through the samtools flagstat output. I know the mapped reads line is probably the first thing to look at but I was wondering what other lines are worth paying attention to.

For example, what would a high number of singletons or a low percentage of properly paired reads usually indicate? Please correct me if I am wrong, but I was thinking that a low properly paired rate could sometimes be a sign of structural variants.

Are there any values in the flagstat output that would make you think something is not right and I should go back and check the earlier step?

• 0 views • link
written by ptalebic • ptalebic

I suggest you use samtools stats for deep drive into the report after flagstat is done. This gives a more robust stats within your reads

Prash

• 0 views • link
written by Prash • Prash

Log in to answer this question.

More posts like this