You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We should add a new step to the workflow that takes the filtered SCE object and applies normalization and dimension reduction (UMAP), storing those results in the SCE that is finally published.
This step should immediately precede the QC report.
We should follow the same code steps that we use in https://github.com/AlexsLemonade/scpca-downstream-analyses, but I expect we will want to combine any additional filtering, normalization, and the UMAP calcualtion into a single script so it can run all together.
One question is whether we want to include filtering as specified by miQC, and any additional criteria. I had the thought that we could do this in a "soft" filter. In other words, we would do so before calculating UMAP, and then merge the calculated UMAP matrix in with the full data set, setting values to NA for cells that were filtered out. This allows us to keep the QC template as is, but still include UMAP plots for only those cells that pass additional filtering (avoiding a cluster of high-mito cells, for example)
The text was updated successfully, but these errors were encountered:
allyhawkins
changed the title
Add normalization and dimension reduction to workflow
Add script to perform normalization and dimension reduction
Aug 29, 2022
We should add a new step to the workflow that takes the filtered SCE object and applies normalization and dimension reduction (UMAP), storing those results in the SCE that is finally published.
This step should immediately precede the QC report.
We should follow the same code steps that we use in https://github.com/AlexsLemonade/scpca-downstream-analyses, but I expect we will want to combine any additional filtering, normalization, and the UMAP calcualtion into a single script so it can run all together.
One question is whether we want to include filtering as specified by miQC, and any additional criteria. I had the thought that we could do this in a "soft" filter. In other words, we would do so before calculating UMAP, and then merge the calculated UMAP matrix in with the full data set, setting values to NA for cells that were filtered out. This allows us to keep the QC template as is, but still include UMAP plots for only those cells that pass additional filtering (avoiding a cluster of high-mito cells, for example)
The text was updated successfully, but these errors were encountered: