Absence of eQTLs for some genes about mixrf HOT 5 CLOSED

Thanks for reporting this! The simplest workaround may be:

1. replace the elements of eqtl.lis for genes without eQTLs with a fake intercept term, say intercept;
2. add a column of 1 with name intercept to snp.dat.

from mixrf.

By adding a column of 1, do you mean that all the elements of the column intercept in snp.dat should be 1?

from mixrf.

Yes, this would not introduce new information but get rid of the error message.

from mixrf.

It works now, but it takes plenty of time if there are more number of genes. The dimensions of my dataset is 175 x 4812 x 4 and it never completes. Is this usual?

from mixrf.

As stated in the paper, we suggest parallel computation. Based on the following statement from the paper, I guess you may need to run about 24 hours for your data using a single node. You can get a more accurate estimate of the time by imputing a single gene (say t1) and the total time estimate is t1 x 4812.

With parallel computing, imputing 10,000 genes in 9 tissue types from about 150 individuals and obtaining the 10-fold CV-based measures of imputation quality could be completed within 30 hr with a 40-node cluster (3.0 GHz Intel Xeon E7 processor) and 16.5 GB of memory.

from mixrf.