[Jaje] Fwd: UKB RAP Q&A session

[Julia Ramirez]

For info...Hubo una sesión específica de RAP ayer antes del congreso y 
fue Helen, nos manda sus notas.

Julia



-------- Forwarded Message --------
Subject: 	Re: UKB RAP Q&A session
Date: 	Tue, 10 Dec 2024 17:28:20 +0100
From: 	William Young <w.young en qmul.ac.uk>
Reply-To: 	pbmlab <pbmlab en qmul.ac.uk>
To: 	pbmlab <pbmlab en qmul.ac.uk>



Thanks Helen for sharing.

BW

Will


==========================================
/Dr William J Young MRCP, PhD/
/NIHR Clinical Lecturer, Queen Mary University of London/
///Electrophysiology Cardiology Registrar, Barts Heart Centre//
/
------------------------------------------------------------------------
*From:* pbmlab-request en qmul.ac.uk <pbmlab-request en qmul.ac.uk> on behalf 
of Helen Warren <h.r.warren en qmul.ac.uk>
*Sent:* Tuesday, December 10, 2024 4:16:45 PM
*To:* pbmlab <pbmlab en qmul.ac.uk>
*Subject:* UKB RAP Q&A session
Dear all,

Sharing here my notes from this morning's pre-conference session, as I 
was the only one who attended...

(Hope my quick notes make some sense to others...)

Best wishes,
Helen

UKB Dec 2024 Conference


## UKB-RAP Q&A session:

  * Mandatory training!!
  * UKB GitHub has lots of notebooks with example scripts, which also
    state recommended instances, to give you an idea of run-time & cost etc
  * Parallel computing always better
  * Better/cheaper to use bespoke apps
  * Lots of training & videos available
  * If ever we need more compute power for large-scale analyses, we can
    email the UKB dnanexus support team, and request more, they’re happy
    to help
  * Recently published a paper on policy guidance for how best to use AI
    on the UKB dataset
  * Use UKB community forum
  * eg Q about working group for AI & ML methods in UKB
  * Data Showcase can’t be used to cross-reference or cross-tabulate
    between different data-fields. For data anonymity. So, if we have a
    specific query,
  * One of the simple examples for running a GWAS, is for height; a more
    complex one is for BP! And helps with categorisation of different
    sources of different BP data field measurements
  * Simple example is using PLINK (which is ok for small N datasets)
  * Then also a step-by-step tutorial for running a GWAS using REGENIE
    (which is better for large N analyses)
  * Q from UKB team, whether researchers would find it helpful to have a
    synthetic dataset to use for tutorials & training >> I said YES!

------------ próxima parte ------------
Se ha borrado un adjunto en formato HTML...
URL: <http://listas.unizar.es/pipermail/jaje/attachments/20241211/a2c8b437/attachment.html>