Upload and display of DMS data from very closely-related experiments

[MaximilianStammnitz]

Hi,

Just about to upload my first ever data to MaveDB (yay!)

After reading the Key Concepts description on Experiment Sets vs Experiments vs Score Sets, I was left wondering about one thing: how should we deal with / display those kind of experiments which were performed on the same library and with the same (or very similar) setup, but which feature small parametrisation differences?

In our specific case it is a dose response series of the exact same assay setting and DMS library separately measured across 12 drug concentrations; to look at small changes in fitness, compare mutant EC50s and similar things, it might be helpful if resulting heatmaps are visualised jointly, e.g. stacked on top of each other. Perhaps a similar reasoning may also apply to others' time course snapshot data, different bins of a FACS and the like?

My understanding is that such scores should all be supplied under the same Experiment Set, but then each drug concentration (or time point, FACS bin, etc.) form their own, separate experiment. Could you please let me know if this is the intended standard? Or are there other opinions on how to pack these kind of closely linked datasets together even a bit more?

In any case – just wanted to say MaveDB is an amazing resource! And sincere apologies in case I have missed something crucial in the documentation.

Thanks a lot,
Max

[afrubin]

Thanks @MaximilianStammnitz for these questions and for uploading your data!

In our specific case it is a dose response series of the exact same assay setting and DMS library separately measured across 12 drug concentrations; to look at small changes in fitness, compare mutant EC50s and similar things, it might be helpful if resulting heatmaps are visualised jointly, e.g. stacked on top of each other. Perhaps a similar reasoning may also apply to others' time course snapshot data, different bins of a FACS and the like?

My understanding is that such scores should all be supplied under the same Experiment Set, but then each drug concentration (or time point, FACS bin, etc.) form their own, separate experiment. Could you please let me know if this is the intended standard? Or are there other opinions on how to pack these kind of closely linked datasets together even a bit more?

Yes, this is the correct interpretation for different drug treatments. Each assay should have its own experiment entry, even if the conditions are similar to another entry.

I agree that this use case is a bit clumsy (for the uploader and for the user who is trying to interpret the data). We're currently working on improving features for comparing across datasets, and the stacked heatmap for multiple datasets on the same target is a good idea. Maybe a heatmap of differences would work well also.

Let us know if you'd like support uploading the data via the API.