Hello,
I’m attempting to use these linear models again and receiving the error message
“Error
Possible causes of error (last one being the most relevant):
Unknown Error Occurred”
when I try to use a continuous variable as the primary metadata. I’m receiving this same error when I put any variable in the covariate or blocking factor as well…
Unless otherwise explicitly stated, I don’t think we support continuous variable as primary data at this stage.
That’s odd because I used this analysis already in Metaboanalyst 5.0 for several tissues a year ago, and got results.
We had this conversation here: Error in Linear models with covariate adjustments - MetaboAnalyst - OmicsForum
And after some time it was fixed and all analyses worked. Why has it been removed?
This is no doubt that we would like support this and many other features. It is just a matter of time and resources. We do officially support those that are formally documented in our published work, as they are directly related to our research work.
You can see some other similar requests from this recent posts and my responses.
Hello Jeff,
I have tried to do this analysis again with categorical data (transforming the continuous) and still receiving the same error - irrespective of if any covariates are included or which contrast group are selected.
I’m thinking there’s a error with the analysis not the data…
Here is an example using the " LC-MS peak intensity data collected from only wild type Arabidopsis thaliana during a wounding time course (four time points). Please refer to (Meinicke P. et al) for more information" time series only data. Log-transformed only.
Showing near all features significant and a perfect linear correlation between p-values. It cannot plot any selected feature either.
I’m happy to share my data to help trouble shoot but this example data suggests somethings not playing right in the code…
Yes, I can reproduce the issue - it is related the plot function for this type of data (time-series only). The statistical analysis is correct. The linear modelling gives 111 significant features, while the conventional two-way ANOVA gives 110 - so it is the characteristic of this particular data.
The server will be updated in the weekend.
This topic was automatically closed after 3 days. New replies are no longer allowed.
