[ensembl-dev] motif features - what's a "good" score?

Ilias Lavidas ilavidas at ebi.ac.uk
Tue Nov 12 13:16:41 GMT 2019

Hello Julie,
Defining a specific cutoff score is not very easy, as this is something 
that varies across cell types.
Since we are focused on potential binding within our predicted 
regulatory elements, we picked a non strict p-value of 0.01 so that we 
wouldn't miss potentially interesting sites.

There is a subset of our motif features that we label as 'experimentally 
verified' and could potentially be of higher value to you.
This subset contains features that fully overlap with a relevant 
ChIP-Seq peak.
At the moment, there is no way to easily retrieve this subset in gff 
format, but we'd be happy to offer this option if this seems useful to you.


On 11/11/2019 17:11, Julie Sullivan wrote:
> Hello! I read the blog post 
> <http://www.ensembl.info/2018/10/15/new-ensembl-motif-features/> but 
> have some questions about the associated data file 
> Homo_sapiens.GRCh38.motif_features.gff.gz 
> <http://ftp.ensembl.org/pub/release-98/regulation/homo_sapiens/MotifFeatures/Homo_sapiens.GRCh38.motif_features.gff.gz>.
> Do you know what p-value was used in MOODS to generate motif feature 
> files? The MOODS docs give "|-p 0.0001" as the example p-value, was 
> that what was used?
> |
> |
> |
> Specifically what I am really interested in how I can filter using the 
> score in the motif features GFF file. The scores seem to range from 
> -55 to 24 (or so) -- is there a associated confidence range, e.g. 
> "high confidence" = 20-24? Or a cutoff score? If that makes sense?
> Or maybe you have more information as to how I can interpret this 
> score? I read the MOODS paper which did not describe this score, and 
> only linked to another paper which I do not have access to.
> Thanks so much! :)
> Julie
> _______________________________________________
> Dev mailing list    Dev at ensembl.org
> Posting guidelines and subscribe/unsubscribe info: https://lists.ensembl.org/mailman/listinfo/dev_ensembl.org
> Ensembl Blog: http://www.ensembl.info/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ensembl.org/pipermail/dev_ensembl.org/attachments/20191112/4572e565/attachment.html>

More information about the Dev mailing list