inst/www/helpfiles/filtering_parameters.md

In ChromSCape: Analysis of single-cell epigenomics datasets with a Shiny App

Filtering parameters

Due to the current technical limitations of single-cell epigenomic technologies, the signal is sparse, some cells are lowly covered and potential doublets might be present in droplet-based technologies. To face these issues stringent filtering is necessary and the dataset can be filtered using 3 different parameters :

• Minimum coverage per cell - represented by the green line on the plot. Setting this filter will remove lowly covered cells. If you observe two peaks in the distribution, you can manually set this threshold in between the two peaks as the first peak might be considered as empty droplets with contaminant DNA.
• Upper percentile of cells to remove (potential doublets) - represented by the red line on the plot. Cells with an outlier number of reads should be discarded as there is a non-zero percentage of droplets containing 2 cells or more (doublets).
• Minimum percentage of cells to support a window - Regions not supported by this percentage of cells (with coverage greater than 1,000 reads) are filtered out. These regions might not be relevant to the analysis. The threshold at 1% is recommended for genomic bins feature, but might be set to 0 if reads were counted on predefined regions / features of intereset.

It is recommended to try out multiple combination of parameters to see what is the impact on clustering & differential analysis.

ChromSCape documentation built on Nov. 8, 2020, 6:57 p.m.