Comparing and assessing the quality of gene annotations between two mouse strain genomes

Question

[TheWonderer]

I've been working on determining the quality of two strain specific assemblies. Thus far, I've attempted to use a few different software (gffcompare, UCSC liftover, and g2gtools) but they don't seem to be best for handling the problem. I've tried to perform a liftover but the chain files are unavailable and even tried to make my own using g2gtools but it had some significant issues. I have considered using BUSCO but I'm not certain the evaluation will provide me details associated with the gene annotations. Please let me know if you have any suggestions, thank you!

Answer 1

[TheWonderer]

I decided to use Liftoff for the purposes of answering this question.

Here's the GitHub link: https://github.com/agshumate/Liftoff

It produces some good metrics for identifying differences/quality in annotations. The tools doesn't require a chain file which makes it easy to use for my request related to mouse strains. I highly recommend making a conda environment when using Liftoff which I think is in the documentation.

Answer 2

[Istvan Albert]

your post slightly conflates two different issues: what is the difference and which annotation is better

finding differences is not so hard, gffcompare for example or simple alignments can do it

finding out, overall which assembly is better is far more difficult as both builds might have strengths and weaknesses and those tradeoffs need to be scored in some way

scientists have been producing genome builds over extended periods, years - a new, better build takes a long time to hone

One way to assess the quality of annotations is to match the genome builds to sequencing reads, transcript sequences or protein sequences deposited in various repositories.