Question About Ensembl Biomart Tables

Question

[User 9996]

Had a few questions about Biomart tables from Ensembl:

1. When I download the tables for hg19 and check only unique results, I get about 150,000 entries -- why is that? Shouldn't it be more like 25,000 or 30,000?

2. When I try to check more than two or three entries in the "EXPRESSION" section of the "Attributes", it errors that I checked too many entries -- why does that happen?

thanks.

[Joachim]

Download which tables? Are you querying the BioMart for Ensembl Gene IDs and Ensembl Transcript IDs?

Answer 1

[Prateek]

1. Remember, there are multiple (alternative) transcripts (ENSTs) per gene. So there are only 52580 unique genes (ENSG...) . Many of the transcripts are predicted proteins (NM vs NP in refseq lingo) for which you would not find ENSP... identifiers and protein sequences. [?][?]
2. From the expression section, 3 is the max number of attributes you can select at a time (as it says in the form) - maybe you are trying to check more than three.

Answer 2

[Joachim]

It seems that you are querying the attributes 'Ensembl Gene ID' and 'Ensembl Transcript ID' in the 'Homo Sapiens genes' dataset of the 'Ensembl Genes 60' BioMart. As Prateek says, this query returns 52,580 Ensembl Gene IDs and additionally their respective Ensembl Transcript IDs. Since a gene can produce multiple transcripts, you get 157,601 pairs of gene/transcript IDs in total.

You probably expect only 20,000 -- 30,000 genes, because you are only considering the known protein-coding genes. In Ensembl, you have more classes of genes recorded though, where you can get an overview of the stored data here (look for 'Gene counts'): http://www.ensembl.org/Homo_sapiens/Info/StatsTable?db=core

The restriction on attributes in BioMart is enforced by the person who configured the BioMart in order to prevent a single person to bring the database to its knees. If you install BioMart locally, you can of course disable said restriction and run arbitrary complex queries on your own database. A slightly outdated version of creating your own local mart can be found here: http://bergmanlab.smith.man.ac.uk/?p=35