search

bvilhjal / ldpred

MIT License
96 stars 57 forks source link

using of three datasets: reference panel, summary statistics panel and validation dataset #73

Closed wym0072003 closed 5 years ago

wym0072003 commented 5 years ago

Hi,

I am sorry for bothering you again. Just a little bit confused with the usage of three datasets: reference panel, summary statistics panel and validation dataset.

It looks the validation and reference panel can be the same dataset. How about the summary statistic dataset, could it be the same the reference panel? That is to say, reference panel, summary stat dataset and validation dateset is the same data. Thank you in advance.

best,

Yiming

bnj50 commented 5 years ago

also how to assign different columns when we have more than 1 summary data? and what is the --ssf command line (--ssf 1.txt, 2 .txt)?

bvilhjal commented 5 years ago

Thanks for these comments, and I apologize the slow reply (was on holiday in April). If you have access to the inidividual-level data (on which the summary stats are generated), then I recommend using other methods, e.g. BOLT-LMM or penalised regressions. LDpred is good for training on summary statistics when you do not have access to individual-level data.

Best.