This repository contains codes for running ALRA in R V体育ios版. The only prerequisite for ALRA is installation of the randomized SVD package RSVD which can be installed as install.packages('rsvd').

The functions now have a flag use.mkl for users who have installed rpca-mkl, which allows for dramatic speedups over the default rpca-based version. Note that rpca-mkl is still under development and is not on CRAN, so it is not a required package, but if users have already installed it then they can use it by setting this flag to True VSports最新版本.

ALRA can be installed as follows:

install.packages("devtools")
devtools::install_github("KlugerLab/ALRA")

Please be sure to pass ALRA a matrix where the cells are rows and genes are columns.

ALRA can be used as follows:

library(ALRA)

# Let A_norm be a normalized expression matrix where cells are rows and genes are columns.
# We use library and log normalization, but other approaches may also work well.
result.completed <- alra(A_norm)
A_norm_completed <- result.completed[[3]]

library(ALRA)

data("b_nk_example")
data("labels_example")

Library and log normalize the data

A_norm <- normalize_data(b_nk_example)

Choose k.

k_choice <- choose_k(A_norm)

For the results in the paper, automatically chosen k worked quite well, but in some cases you might want to take a closer look, as we do here. The k is chosen based on the spacings between the singular values, as it can be quite hard to identify the ``beginning of noise'' from just looking at the spectrum itself V体育官网入口.

library(ggplot2)
library(gridExtra)
df <- data.frame(x=1:100,y=k_choice$d)
g1<-ggplot(df,aes(x=x,y=y),) + geom_point(size=1)  + geom_line(size=0.5)+ geom_vline(xintercept=k_choice$k)   + theme( axis.title.x=element_blank() ) + scale_x_continuous(breaks=seq(10,100,10)) + ylab('s_i') + ggtitle('Singular values')
df <- data.frame(x=2:100,y=diff(k_choice$d))[3:99,]
g2<-ggplot(df,aes(x=x,y=y),) + geom_point(size=1)  + geom_line(size=0.5)+ geom_vline(xintercept=k_choice$k+1)   + theme(axis.title.x=element_blank() ) + scale_x_continuous(breaks=seq(10,100,10)) + ylab('s_{i} - s_{i-1}') + ggtitle('Singular value spacings')
grid.arrange(g1,g2,nrow=1)

A_norm_completed <- alra(A_norm,k=k_choice$k)[[3]]

Check the completion. Note that the results improve when using the entire dataset (see the ALRA-paper repo for those codes), as opposed to this subset VSports在线直播.

• NCAM1 should be expressed in all NK cells, but is only expressed in 4% of of NK cells in the original data.
• CR2 should be expressed in all B cells, but is only expressed in 1% of of B cells in the original data.

print(aggregate(A_norm[,c('NCAM1','CR2')], by=list(" "=labels_example),FUN=function(x) round(c(percent=100*sum(x>0)/length(x)),1)))
print(aggregate(A_norm_completed[,c('NCAM1','CR2')], by=list(" "=labels_example),FUN=function(x) round(c(percent=100*sum(x>0)/length(x)),1)))