Request for annotated isoform dataset with CAGE/polyA validation for ML classifier development

[VladimirOpaits]

Hello ConesaLab team,
I'm Vladimir, a second-year Data Science student at Universitat de València working on developing an ML-based isoform quality scorer. My goal is to build a tool that assigns confidence scores to isoforms to help quickly filter obvious artifacts from large transcriptome datasets. Can you please give me a dataset (any organism/cell line) where I can access SQANTI3 classification file with within_CAGE_peak / within_polyA_site fields populated ?

[pabloati]

Hi Vladimir,
You have some examples of SQANTI3 results and classification tables in our examples repository.

Let me kno if they are useful to you or you would need something else :)

[VladimirOpaits]

Thank you, Pablo! Chr22 is a great starting point. I'm planning to build an external hyperparameter validation study for the RF filter — basically benchmarking whether the default parameters are optimal for different dataset sizes. If you have classification files for more chromosomes, that would be really helpful for scaling the analysis!