How to compare different annotation of transcriptome with each other?

Question

[jack]

Is there any systematic way to compare human transcriptome annotation of different databases like UCSC, RefSeq and Ensemble with each other? I was wondering how much isoforms annotation overlap between different databases. For example for gene A, there is 3 isoform at UCSC, but there is only one isoform at UCSC databases.

Answer 1

[mastal511]

Each of the different databases (Ensembl, NCBI RefSeq, and UCSC), will use different methods for annotation, which they will tell you about somewhere on their website. Different databases will vary from being highly-curated, some in part manually, and others annotated from computational gene predictions. In general, highly-curated would be assumed to be more accurate.

[LLTommy]

This is certainly true that manually curated databases should be better. But still, there is no general way to decide this, right? - e.g. which one of the manual curated databases is 'better' than another one? This is hard to say and I guess most people would say something like 'in my opinion....' or 'judging by my experience...' - wouldn't you agree?

I guess, in the end - this is why you always cite the database, including data/version where you base you results on - 'cause stuff can change ;)

Answer 2

[LLTommy]

If I understand your question correctly, your are dealing with different annotations for different (or maybe the same) ..things (generally spoken). If so, you have to understand where annotations are coming from and that are (different) ontologies. So something like Ensemble does not do their 'own annotations', rather the annotate stuff to a certain ontology term. So if you have an id like for example GO:0000XXX that gives you a hint that it is annotated to the gene ontology.

I am not sure if that helped you but if not, please follow you post up with a question or an example and I shall reply if I am able to.

[jack]

I update the post, it's not about the onotology terms, it's more about which database contain more accurate annotation of human transcriptome.

[LLTommy]

I see. Well, I can not help you with that. I don't know if anyone can say what database has more accurate annotations (or results).... or how to judge which one is in general 'better'. Maybe that just depends on the question and the individual case.... So, I don't know. However, I would be interested in other people's opinion.