high values of the Kolmogorov-Smirnov score for several dataset

[ERIGR]

Hi, I get an extremely high KS score for three of the four datasets I want to integrate (a value of 1 for three datasets among four). Moreover when running the function InPlot, I get thes kind of messages :
1: In dpois(y, mu, log = TRUE) : non-integer x = 0.873301
2: In dpois(y, mu, log = TRUE) : non-integer x = 0.907027
3: In dpois(y, mu, log = TRUE) : non-integer x = 0.935194
4: In dpois(y, mu, log = TRUE) : non-integer x = 0.947587
5: In dpois(y, mu, log = TRUE) : non-integer x = 0.955919

How should I deal with that ? What does it mean ?
Thanks

[yangRISC]

hi ERIGR, for your questions, I have some suggestions.

1. If KS score is very high, I suggest you could use RISC (v1.5.0). In the new version, we add some parameters in “scFilter” function to filter abnormal genes, e.g., “gene.ratio” = 0.02 and “mitochon” = 0.1. Furthermore, you could use higher “min.gene” and “min.cell” to remove poor quality cells, e.g., min.gene = 500, min.cell = 5, min.UMI = 1000, and max.UMI = ??? (based on “FilterPlot” function).
2. About “In dpois(y, mu, log = TRUE) : non-integer x =”, if you get this in warnings when running “InPlot” function, that is the general situation. We employed GLM poisson regression to estimate the data quality, but GLM poisson can only be used for integer counts (this is the original design of the GLM package). Therefore, it warns when we use here.

Feel free to leave a message when you have other comments.