search

zhenin / HDL

High-definition likelihood inference of genetic correlations (HDL)
95 stars 28 forks source link

error #13

Closed fry3682665 closed 3 years ago

fry3682665 commented 3 years ago

image

zhenin commented 3 years ago

This is indeed weird! Are you just testing the example files? I will look into it.

fry3682665 commented 3 years ago

It is not a test file, but my own GWAS Summary file, because in my previous use it seems that the HDL operation can only be performed if I change my file name to the same file name as the test file and remove the test file.

fry3682665 commented 3 years ago

image This is my data. Is there a problem with the data?Another request is that I am also having trouble with this data when doing LDSC analysis. The problem has been commented on the GIthub of LDSC and I really hope to get your advice

zhenin commented 3 years ago

It is not a test file, but my own GWAS Summary file, because in my previous use it seems that the HDL operation can only be performed if I change my file name to the same file name as the test file and remove the test file.

HDL should be smarter :) You should be able to specify your own file name by setting gwas1.df= and gwas1.df=.

image This is my data. Is there a problem with the data?Another request is that I am also having trouble with this data when doing LDSC analysis. The problem has been commented on the GIthub of LDSC and I really hope to get your advice

The only thing I find a bit abnormal according to the screenshot is that most of SNPs are without rsid. How many SNPs you have in the GWAS files? Could you try to run with UKB_imputed_SVD_eigen99_extraction reference panel? This is the most standard panel we recommend our users to use. This panel is a subset of LDSC 1000 Genomes panel. If too many SNPs in this panel miss in your GWAS, then there might be some issue for both HDL and LDSC.

fry3682665 commented 3 years ago

Thank you very much. It seems that the lack of SNPID is my main mistake. I still want to ask you how to add SNPID in GWAS Summary

fry3682665 commented 3 years ago

It is not a test file, but my own GWAS Summary file, because in my previous use it seems that the HDL operation can only be performed if I change my file name to the same file name as the test file and remove the test file.

HDL should be smarter :) You should be able to specify your own file name by setting gwas1.df= and gwas1.df=.

image This is my data. Is there a problem with the data?Another request is that I am also having trouble with this data when doing LDSC analysis. The problem has been commented on the GIthub of LDSC and I really hope to get your advice

The only thing I find a bit abnormal according to the screenshot is that most of SNPs are without rsid. How many SNPs you have in the GWAS files? Could you try to run with UKB_imputed_SVD_eigen99_extraction reference panel? This is the most standard panel we recommend our users to use. This panel is a subset of LDSC 1000 Genomes panel. If too many SNPs in this panel miss in your GWAS, then there might be some issue for both HDL and LDSC.

image image I used my own file name, the data format conversion was complete, and the HDL operation didn't work at all I am eager to get your help

fry3682665 commented 3 years ago

It is not a test file, but my own GWAS Summary file, because in my previous use it seems that the HDL operation can only be performed if I change my file name to the same file name as the test file and remove the test file.

HDL should be smarter :) You should be able to specify your own file name by setting gwas1.df= and gwas1.df=.

image This is my data. Is there a problem with the data?Another request is that I am also having trouble with this data when doing LDSC analysis. The problem has been commented on the GIthub of LDSC and I really hope to get your advice

The only thing I find a bit abnormal according to the screenshot is that most of SNPs are without rsid. How many SNPs you have in the GWAS files? Could you try to run with UKB_imputed_SVD_eigen99_extraction reference panel? This is the most standard panel we recommend our users to use. This panel is a subset of LDSC 1000 Genomes panel. If too many SNPs in this panel miss in your GWAS, then there might be some issue for both HDL and LDSC.

image image I used my own file name, the data format conversion was complete, and the HDL operation didn't work at all I am eager to get your help

zhenin commented 3 years ago

The names of arguments for HDL should be: 'gwas1.df' and 'gwas2.df', instead of 'asd.df' and 'ra.df'. You can change them and try it again.