RNAseq data within seqr

[bellmartin]

Hi seqr team,
I would be grateful if you could let me know when the new features for the integration of RNA-seq data within seqr will be made available for groups with their own seqr instance ?

We have our own pipelines using Fraser ( and now testing Fraser2) as well as outrider and are keen to have these RNA-seq results imported into our own seqr instance. It would be very helpful if you could let us know what format the data (Fraser & outrider results ) should have for ingesting into seqr, so we can begin formatting our data appropriately. Are you planning on releasing your own pipelines for RNA-seq analysis ?

We are happy to be used as a testing site if there are any features you would like feedback on. We have a reasonable cohort of RNAseq samples ( ~85 mRNA samples, majority one tissue type, 4 tissue types in total) with corresponding WGS samples.

Thanks in advance for these new features!

Kind regards,
Katrina

[hanars]

@bw2 has developed the pipeline we use for RNA calling, and can speak to when we anticipate that being available

[bw2]

@bellmartin , seqr supports 4 RNA-seq related data types:

gene expression (TPM)
gene expression outliers (such as from OUTRIDER)
splice junction outliers (such as from FRASER)
IGV.js splice junction tracks (igvteam/igv.js#1019)

@hanars are all of these ready for external use, and what's the best way to document the input formats?

[hanars]

splice junction outliers are not ready quite yet, we should hold off until we have FRASER2 available in our own seqr instance and have finished any tweaks to the data model or the UI, but everything else is ready

[bellmartin]

Thanks @hanars
Our current seqr version is seqr v1.0-88900306, with our code base last merged with your master as of 2023-06-24 (commit c200fdc). We have Outlier, TPM and SpliceOutlier available to select as data types in this version of seqr. Is this still to be updated with new features for these 3 data types or is this fully enabled as is ? Thanks for your help.

[hanars]

SpliceOutlier has not yet been fully tested by our team, so I might hold off on that if I were you until we can iron out any issues that may still arise with it, although if you do want to try it out it theoretically should work. Outlier and TPM should work fine on the version you are running

[bw2]

As far as pipelines, we've implemented FRASER and OUTRIDER on the Hail Batch cloud compute environment. I'm not sure how useful this is for external groups. Also, the process of generating tables for upload into seqr requires annotating the RNA-seq samples with their seqr project and sample id. Unfortunately that step relies on systems that are not externally accessible.

[hanars]

I think as a lightweight MVP for this, we can document what the file format we expect is, and how to call the data, and include a note that groups are responsible for manually adding the project ID and sample ID.

[bw2]

Here's a summary of columns in the RNA-seq TSV tables that data managers can upload via: https://seqr.broadinstitute.org/data_management/rna_seq

Normalized gene expression table (TPMs from STAR or RNAseqQC):

Gene expression outliers (from a tool like OUTRIDER)

Splicing outliers (from FRASER or similar tools)

[hanars]

@bw2 can you also share (at least minimal) documentation for how the data was generated?

[bellmartin]

Thanks @bw2 that is really helpful! I have a few follow up questions if you don't mind.

• For the tissue type, are there specific tissue type names that are accepted or is this an open field ?

• For outrider and fraser data, I assume you load the whole data set for plotting purposes? Not just the top hits ie adj.p value <1 . Correct?

• For the ensembl geneID's, have you encountered any issues with different versions of gene annotation within seqr ?
  Thanks

[bw2]

The output of FRASER and OUTRIDER directly map to these columns.
In the splicing outlier table the rareDiseaseSamplesWithJunction and rareDiseaseSamplesTotal are places to optionally add extra info about how frequently a junction is seen in other samples in the cohort and/or GTEx. These should be added to the FRASER output table or left blank.

[bw2]

I currently get the gene expression table (TPMs) by combining RNAseqQC2 per-sample output tables.

[AlessandroLas]

Hi all,
Also for me it is unclear if there are specific tissue type names that are accepted.
I tried with "blood", "whole blood" and the UBERON code for the whole blood but it won't work.

Many thanks in advance.
Alessandro.