Topic 1 Single_Cell_analysis（2）

Part 1 Scissor——logistic regression

1. download bulk-dataset and phenotype（TP53mutant As phenotype）

load(url(paste0(location, 'TCGA_LUAD_exp2.RData')))
load(url(paste0(location, 'TCGA_LUAD_TP53_mutation.RData')))

table(TP53_mutation)

share 497 individual cells, among 255 Yes mutant, The rest 243 Yes wild-type（mutant Set to 1,wild-type Set to 0）

2.Scissor Select cells with information （informative cells）

# Use scissor Select signal cells （TP53 mutant==1 wild-ttype==0）family=binormal
phenotype<-TP53_mutation
tag <- c("wild-type", "TP53 mutant")
infos3<-Scissor(bulk_dataset,sc_dataset,phenotype,tag=tag,alpha=NULL,family = "binomial",Save_file="Scissor_LUAD_TP53_mutation.RData")

No other parameter values are set here , Final customization ！

It can be seen from the above figure , When alpha=0.2 when , Identify to 414 individual mutant cell and 320 individual wild-type cell, Will identify cell With all cells present in picture

Scissor_select<-rep(0,ncol(sc_dataset))
names(Scissor_select)<-colnames(sc_dataset)
Scissor_select[infos3$Scissor_pos]<- 1
Scissor_select[infos3$Scissor_neg]<- 2
sc_dataset<-AddMetaData(sc_dataset,metadata = Scissor_select,col.name = "scissor")
DimPlot(sc_dataset,reduction = 'umap',group.by = 'scissor',cols = c('grey','indianred1','royalblue'),pt.size = 1.2,order = c(2,1))

Sissor_pos： To express with mutant Closely related cell（ Because in the process of setting mutant by 1）
Scissor_neg： Express wild-type Closely related cell（ Set during setup wild-type by 0）

Part 2 Scissor Interpretation of the results

Scissor_postive Cells and Scissor_negative Cells are associated with specific phenotypes （ Such as TP53mutantion、survival etc. ） Highly relevant Scissor Selective cell .Scissor The association between selected cells and phenotypes depends on the model used , Should be interpreted in a particular context . about Linear regression and classification model , Initial value of parameter phenotype Will affect the interpretation .

for example , In the above use TP53 In the application of mutation state , If TP53 The mutant is encoded as 1, The wild type is 0, be Scissor+ Cells will interact with TP53 Mutants are associated , and Scissor- The cells will be associated with the wild type . If the encoding is reversed phenotype For both phenotypes ,Scissor+ Cells and Scissor- The cellular interpretation changes accordingly . about Cox Return to ,Scissor+ Cells are associated with poor survival , and Scissor- Cells are associated with good survival .

Scissor Cells can be linked to phenotypes , This connection is a relative concept between phenotypes . in other words ,**Scissor Which phenotype is more likely to be associated with one cell than with another phenotype .** Considering that there may be a negative correlation between single cell and somatic samples , We can divide a cell into three categories （ surface 1） To further explain a cell ：** If the average correlation between a cell and all somatic samples is greater than 0, And the positive correlation number is greater than the negative correlation number , This cell is more similar to the related phenotype ; If the average value of a cell correlation is less than 0 And the number of negative correlations is greater than the number of positive correlations , The cell should be interpreted as more dissimilar to other phenotypes ; otherwise , The association of this cell with the phenotype is uncertain .** in the majority of cases , The negative correlation value is very small , The identified cells belong to “ More similar ” Categories .

notes ：Development

1.label

Through all the data Scissor Data can only be processed into Scissor_pos and Scissor_neg Two label——> Multi label

2.model of selected cell and phenotype

Scissor Selected positive and negative cells , According to your own model stay cell and phenotype Selected in the association of （ Context specific selection ） check ： Relevance selection

Question

1. the relationship of single-cell & sample

There may be a negative correlation between single cell and somatic samples ??