Failed to bind results due to missing column
Closed this issue · 6 comments
Hi,
The column Stat
should be included in the dataframe here
Lines 127 to 131 in 04eec91
Otherwise the results binding in
lmmTestAllRegions
will not work due to different number of columns, because you extract 4 columns from lmer test results (Estimate
, StdErr
, Stat
and pValue
), while there are only 3 columns created when the lmer test failed coMethDMR/R/lmmTestAllRegions.R
Line 212 in 04eec91
I suggest to add this to fix the rbind problem :
if (length(results_ls) > 0) {
# outDF <- do.call(rbind, results_ls)
outDF <- data.table::rbindlist(results_ls, fill = TRUE) ## --here fixed
outDF$FDR <- p.adjust(outDF$pValue, method = "fdr")
outDF$region_names <- names(region_ls) ## --here to keep region names
row.names(outDF) <- NULL
}
Wow! Thank you so much for the help.
Do you want to be authors on the package for the next version? We are going to add support for whole genome data next (see Issue #8), but there's a lot of documentation (and clearly bugs) to fix first. If you're interested, I'll send you a zoom link for some morning my time (which should be afternoon where you are).
Hello,
Thank you for your return!
With pleasure, I can continue to give you feedback because my colleagues and I are currently using your coMethDMR package. I can continue to give you ideas for improvements as well as fixed bugs that we will see.
I have never worked on the whole exom yet, but it will be interesting!
Speaking with @Ning-L , I understood that she was interested to better understand why some regions had failed in lmer test.
We will look, if it is possible, to add a piece of code like try(tools::assertCondition()), silent = TRUE)
or tryCatch
to capture the error and return it to the result tables within the dedicated row (region). I will let you know if it works well and if it helps.
Wow! Thank you so much for the help. Do you want to be authors on the package for the next version? We are going to add support for whole genome data next (see Issue #8), but there's a lot of documentation (and clearly bugs) to fix first. If you're interested, I'll send you a zoom link for some morning my time (which should be afternoon where you are).
Glad to contribute.
Please email me the zoom link a few days before the time that suits you so I can pre-schedule my work ^^
@mboissel and @Ning-L, can you please send me an email from your preferred email accounts? I'm gabriel.odom@fiu.edu
Also, I'm adding you both as package contributors for now :)