Comments on PR #111 + sharing trick we used for overlapping spots from overlapping capture areas

[lcolladotor]

Hi BayesSpace authors,

I saw PR #111 which seems interesting. I saw that you are changing some names, like row to array_row and stuff like that. I think that will be important to announce, as we'll have to update some of our code as users to match these changes.

Also, does the sanity check code at 86de359#diff-2de0a6ab966efd85915ee3fa43a1f14c41edbde37cb62d758f551b8296b5febcR248-R260 affect the current joint clustering strategy listed at https://www.ezstatconsulting.com/BayesSpace/articles/joint_clustering.html#clustering-1? The one where you make an artificial array/row grid. It also doesn't seem to be the best practice (and if you ran BiocCheck you'd see this) to have a message() call with the word warning. If you consider this an actual warning, then I'd recommend using warning("some message", call. = FALSE).

Finally, I'd like to share with you a trick that @Nick-Eagles and I documented at https://research.libd.org/visiumStitched/articles/visiumStitched.html#defining-array-coordinates. Basically, we have cases where we can have overlapping data and assign 2 or more spots to the same array row/col hex grid locations, which then allows BayesSpace (and other tools using this hex grid info) to run spatial clustering using data from neighbors (and their overlapping spots). For more details, check also https://doi.org/10.1101/2024.08.08.607222.

Best,
Leo

[lcolladotor]

Hmm, is see at https://github.com/edward130603/BayesSpace/pull/111/files and in particular at https://github.com/edward130603/BayesSpace/pull/111/files#diff-2de0a6ab966efd85915ee3fa43a1f14c41edbde37cb62d758f551b8296b5febcR95 that you plan on adding the scale factors to the object. But this won't work for joint clustering, right? Since each capture area can have different scale factors.

Though it doesn't look like metadata(sce)$BayesSpace.data$scalef is being used anywhere (from searching scalef at https://github.com/edward130603/BayesSpace/pull/111/files).