Hey there,
First of all, the patchseq tool has been an absolute life saver so thanks for making it accessible.
I was wondering if there was a good way to calculate the NMS score for multiple ‘ON classes’ at once? For instance, if patching from both astrocytes and neurons, would it be possible to calculate the contamination score from non-astrocyte subclasses in the case of astrocytes cells and also calculate the contamination score from non-neuronal subclasses for neuronal cells? If so what would it look like - would it be best to go through the pipeline twice, defining different OFF-targets each time? Currently, my astrocytes are not getting mapped since they get filtered out and obviously have poor NMS scores.
Thanks for your help!
Luc
Hi Luc,
Glad you are finding the tool useful! I think your best bet will be to set to off target all non-neuronal types EXCEPT astrocytes. This will likely involve creating a new column in your metadata, similar to what was done here. Similarly if you include astrocytes as part of your on target subclasses, then you should automatically have NMS calculated for the relevant type (astrocyte or neuronal type).
The second option would be to first divide astrocytes from neurons (either manually or computationally), and then to perform separate analyses on both data sets (neurons and astrocytes) using different taxonomy subsets. I don’t think it makes sense to run the analysis twice on all data sets.
Finally, I want to note that NMS score has been tested on neocortical samples and is generally reliable, but may or may not work as well in other brain/body structures. It is worthwhile to use as a QC score, but be especially cautious if trying to use as a hard filter.
Best,
Jeremy
Hey Jeremy,
Brilliant - option 1 did the trick, I now have the NMS and mapping for both cell classes. Super happy thank you! I’ll keep an eye out on whether or not the NMS cutoff makes sense, thanks for the input.
All the best,
Luc