chenweng1991 / ezsinglecell Goto Github PK

Seurat normalization:

Normalize to total umi

After removing unwanted cells from the dataset, the next step is to normalize the data. By default, we employ a global-scaling normalization method “LogNormalize” that normalizes the gene expression measurements for each cell by the total expression, multiplies this by a scale factor (10,000 by default), and log-transforms the result. This is stored in XX@data

Then Regress out

our single cell dataset likely contains ‘uninteresting’ sources of variation. This could include not only technical noise, but batch effects, or even biological sources of variation (cell cycle stage). As suggested in Buettner et al, NBT, 2015, regressing these signals out of the analysis can improve downstream dimensionality reduction and clustering. To mitigate the effect of these signals, Seurat constructs linear models to predict gene expression based on user-defined variables. The scaled z-scored residuals of these models are stored in the scale.data slot, and are used for dimensionality reduction and clustering.
This is stored at [email protected]

DEseq(DEseq2) Normalization

median-of-ratios method, which they claim this perform better than directly using total UMI in cases with some extremely expressed genes. check out their paper. https://genomebiology.biomedcentral.com/articles/10.1186/gb-2010-11-10-r106
BTW, DEseq2 and other tools made a compromise in estimating parameters because of limited number of samples, which is not a problem is single cell study.

Deconvolution method, which we used in DE study .

https://genomebiology.biomedcentral.com/articles/10.1186/s13059-016-0947-7#Equ1

It is essential to not forget what is the algorithm of clustering in Seurat.
Basically, it is a graph-based community detection method SLM(smart local moving). There are other SLM packages out there, eg. https://github.com/chen198328/slm and https://github.com/deepminder/SLM4J

There are also other community detection algorithm, such as this https://github.com/CWTSLeiden/networkanalysis, which as claimed better then SLM.

It is probably helpful to try out SCENIC or their principle to

• Study regulatory network of differentially expressed genes
• Coexpressed gene module on TreCCA tree

https://www.nature.com/articles/nmeth.4463