Last updated: 2022-09-15
Checks: 2 0
Knit directory: chromap_vs_cellranger_scATAC_exploration_10x/
This reproducible R Markdown analysis was created with workflowr (version 1.7.0). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.
Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.
Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.
The results in this page were generated with repository version 10fdcb0. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.
Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:
working directory clean
Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.
These are the previous versions of the repository in which changes were made to the R Markdown (analysis/Conclusions.Rmd) and HTML (docs/Conclusions.html) files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view the files as they were in that past version.
| File | Version | Author | Date | Message |
|---|---|---|---|---|
| Rmd | 10fdcb0 | jeremymsimon | 2022-09-14 | Initial commit |
This is the easiest comparison to make and conclusion to draw.
The Sankey plot and NMI values indicate the clustering is quite similar despite the differences in upstream data processing. Further, the Gene Activity values (which are peak agnostic) were extremely highly correlated (nearly all >0.9). This, combined with the fact that the chromap and cellranger counts underneath a unified set of features were also very highly correlated (nearly all >0.9) strongly corroborates the data provided in the chromap paper. This isn’t surprising, but provides additional real-world evidence that the alignments performed by cellranger vs those by chromap are nearly identical.
The peak calls themselves were also very similar (~85% overlap), therefore these data strongly suggest that any differences in downstream clustering are likely attributable to the subtle differences in peaks called.
Nonetheless, these data serve as a proof-of-principle that a chromap-based approach is not only feasible, but can largely capture the same information as cellranger in a tiny fraction of the runtime.
Additionally, these data illustrate something about which I frequently am asked: that integration procedures can preserve cell/sample-specific information. A fear shared by many experimentalists is that if we integrate e.g. treated and untreated cells, and some cell clusters are expected to be treatment specific, that those cell types will be lost in the process. Here, B-lymphoblasts are seemingly different enough from the various cell types present in the PBMC pool, but integration doesn’t force these clusters to overlap. We still see clusters specific to the various PBMC cell types, and the HGMM cell types are both retained and remain separated.
FeatureMatrix can be slow too. Are their equivalent but faster approaches for counting that we could utilize to build our cells x peaks matrices?
Comments
Questions, comments, improvements, and general discussion are all welcome! Start a discussion on github, email me, or find me on Twitter