Closed konkam closed 3 years ago
Thank you for noticing this. The commit c4a3f20 fix the mismatch: I keep the code and change the documentation.
About 'replicate', you are totally right: the new changes imply wrong replicate (every time-series is a replicate in a dataset, whatever are the profile of concentration). This comes from the fact that time-series were previously identified by the couple (concentration, replicate), but now, as we deal also with varying exposure profile were concentration may change in a single time-series, we need a label for every time-series which is 'replicate'. Also, since previously, column 'replicate' was required for pooling 'true replicate', this is now automatized.
The help of the function states:
however the code tests that there are two identical (replicate, time) pairs.
The check for triplets may be more meaningful than the check for pairs, as some users may define replicate rather loosely as "batch of experiments" on a given day, or a given arm of the lab. Therefore they might have replicates with different concentrations.
However this is open to discussion. I think the current definition of replicate in the help of function survData precludes the ambiguity I described above: