MulTiXcan (Multi-Tissue PrediXcan) release

[Alvaro Barbeira]

Dear PrediXcan and MetaXcan community members,

Today we release a major update to the MetaXcan repository that includes the implementation of a new method for multiple-tissue association: MulTiXcan for individual level data, and its summary-level data sibling SMulTiXcan. You can find a preprint manuscript here explaining the details of the method.

These methods yield a gene-level association test that aggregates information across multiple tissues while taking their correlation into account. We hope they will prove useful to your research. Documentation is still a work-in-progress; please let us know if you find it lacking in any respect.

This release also fixes some edge cases in input GWAS parsing, and includes usability improvements.

Best regards,

The PrediXcan & MetaXcan team

[Misha Vysotskiy]

Hi Alvaro

Can you help me resolve this error:

INFO - Processing

Level 7 - Gene: ENSG00000228437.1

INFO - Unexpected error: zero-size array to reduction operation minimum which has no identity

I am running Smultixcan using pgc-scz2 sumstats, Metaxcan output from version downloaded ~6 months ago, gtex-v7 models, and the new snp covariance file from PredictDB. Numpy versions 1.14, scipy 1.0, python 2.7. Default parameters. (Software asked to include at least one cutoff parameter, I used --cutoff_threshold 4).

The gene it's getting stuck on seems to be one modeled in Lung only. 

Thanks,

Misha 

[Alvaro Barbeira]

Hi Misha,

I would advice using --cutoff_condition_number 30 , which is what we are currently using. Please run with --throw parameter, so that if it still fails, I can see a full error trace from the output. I also suggest re-running MetaXcan with the latest version if possible.

It also seems that I wrote it wrong in the command line reference wiki, --cutoff_threshold should not be larger than 1 in general. I'll update the docs. Thanks for the detailed information, I'll debug on my end as soon as possible.

Best,

Alvaro

--
You received this message because you are subscribed to the Google Groups "PrediXcan/MetaXcan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to predixcanmetaxcan+unsubscribe@googlegroups.com.
To post to this group, send email to predixcanmetaxcan@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/predixcanmetaxcan/d89be4fa-83ab-4b11-a3e1-bc186f1773fb%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Misha Vysotskiy]

This fix worked. Thanks!

To unsubscribe from this group and stop receiving emails from it, send an email to predixcanmetax...@googlegroups.com.
To post to this group, send email to predixca...@googlegroups.com.

[Alvaro Barbeira]

Hi Misha,

Glad to hear it works.

I managed to reproduce your issue and fix it in case you want to go back to cutoff threshold.

Best,

Alvaro

To unsubscribe from this group and stop receiving emails from it, send an email to predixcanmetaxcan+unsubscribe@googlegroups.com.
To post to this group, send email to predixcanmetaxcan@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/predixcanmetaxcan/50cfd471-7a96-416b-955d-5fa3c962182d%40googlegroups.com.