library(ggplot2)
library(zeallot)
library(data.table)
library(parallelDist)
library(tidyr)
library(tidyverse)
library(caret)
theme_set(theme_bw())
library(COTAN)
library(stringr)
library(nnet)
options(parallelly.fork.enable = TRUE)
inDir <- file.path("Data/MouseCortexFromLoom/")
#setLoggingLevel(2)
#setLoggingFile(file.path(inDir, "MixingClustersGDI_ForebrainDorsal.log"))
outDir <- file.path("Data/MouseCortexFromLoom/PureClusters/")
if (!file.exists(outDir)) {
dir.create(outDir)
}Real UT clusters
Cl432 E13.5 E15.0
Preamble
428 and 434 e15.0
e15.0 <- readRDS(paste0(inDir,"e15.0_ForebrainDorsal.cotan.RDS"))
cells <- c(rownames(getMetadataCells(e15.0)[getMetadataCells(e15.0)[,"Clusters"] == 428,]),
rownames(getMetadataCells(e15.0)[getMetadataCells(e15.0)[,"Clusters"] == 434,]))
cl428_cl434 <- dropGenesCells(e15.0,cells = getCells(e15.0)[!getCells(e15.0) %in% cells] )
cl428_cl434 <- clean(cl428_cl434)
cl428_cl434 <- proceedToCoex(cl428_cl434)
#saveRDS(cl428_cl434,past0(outDir,"cl428_cl434.cotan.RDS"))cl428_cl434GDI <- calculateGDI(cl428_cl434)
subsetGDIcl428_cl434 <- cl428_cl434GDI[cl428_cl434GDI$sum.raw.norm > 7,]
top.GDI.genes <- rownames(subsetGDIcl428_cl434[order(subsetGDIcl428_cl434$GDI,decreasing = T),])[1:50]
GDIPlot(cl428_cl434,genes = "",GDIIn = cl428_cl434GDI)
data <- getNormalizedData(cl428_cl434)
data <- data[!rowSums(as.matrix(data)) < 1,]
data <- log(data*10000+1)
row_stdev <- apply(data, 1, sd, na.rm=TRUE)
row_stdev <- row_stdev[order(row_stdev,decreasing = T)]
genes.to.keep <- c(names(row_stdev[1:100]),top.GDI.genes)
data.small <- data[rownames(data) %in% genes.to.keep,]
#data <- t(as.matrix(data))
data.small <- t(as.matrix(data.small))
COTAN_Cl.code <- as.numeric(getClusterizationData(cl428_cl434,clName = "original.clusters")[[1]])
COTAN_Cl.code <- COTAN_Cl.code -1
data.small <- cbind(data.small,COTAN_Cl.code)
data.small <- as.data.frame(data.small)
# Split the data into training and test set
set.seed(123)
training.samples <- data.small[,"COTAN_Cl.code"] %>%
createDataPartition(p = 0.8, list = FALSE)
train.data <- data.small[training.samples, ]
test.data <- data.small[-training.samples, ]
head(train.data) Abracl Cdkn1c Cited2 Mllt3 Fam210b
10X74_4_A_1:GTAACGTGCCATAGx 10.128836 0.00000 9.435728 9.435728 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 10.211198 0.00000 10.904327 10.211198 9.518087
10X74_4_A_1:GATTCGGAGGTGAGx 10.518544 9.13233 9.825424 9.132330 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 10.916737 0.00000 9.307371 10.405923 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0.00000 11.194560 0.000000 0.000000
10X74_4_A_1:CTAGGATGCTTGCCx 11.828628 0.00000 10.941336 8.995532 8.995532
Mfap4 Cxcl12 Cdc42ep3 Limch1 Ier2
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 9.435728 0.000000 0.000000 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 9.518087 0.000000 9.518087 0.000000 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 11.078148 9.825424 0.000000 9.132330 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 10.405923 0.000000 10.000473 10.000473 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 9.585177 0.000000 0.000000 0.000000
10X74_4_A_1:CTAGGATGCTTGCCx 10.941336 9.688617 0.000000 8.995532 8.995532
Ptprk Gm29260 Sfrp1 Pantr1 Magi2
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 0.000000 9.435728 10.534287 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 10.211198 9.518087 9.518087 9.518087 9.518087
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 9.825424 9.825424 10.518544 0.000000
10X74_4_A_1:CTGAGCCTGGTAAAx 9.307371 10.000473 0.000000 10.405923 0.000000
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0.000000 9.585177 0.000000 0.000000
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 8.995532 9.688617 10.604871 10.381733
Tmem108 Ddr1 Ezr Sncaip CT025619.1
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 0.000000 9.435728 0.000000 9.435728
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 0.000000 0.000000 9.518087 10.211198
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 0.000000 10.230871 9.825424 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 9.307371 0.000000 0.000000 10.000473 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.000000 10.278290 0.000000 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 8.995532 8.995532 8.995532 0.000000 8.995532
Fabp7 Ctnnd2 Ptn Frmd4a Sorbs2
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 0 10.128836 11.227421 10.534287
10X73_3_A_1:ACGCCGGATCGTAGx 9.518087 0 9.518087 9.518087 10.616651
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 0 0.000000 0.000000 9.825424
10X74_4_A_1:CTGAGCCTGGTAAAx 9.307371 0 10.693598 9.307371 10.000473
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0 10.278290 0.000000 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 9.688617 0 0.000000 0.000000 10.094061
Hes6 Dcc Neurog2 Tcf12 Fam110a
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 9.435728 0.000000 0.000000 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 9.518087 0.000000 0.000000 10.616651 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 9.825424 9.825424 10.741682 9.825424 9.825424
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 10.916737 0.000000 10.000473 0.000000
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0.000000 0.000000 11.194560 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 8.995532 8.995532 10.094061 0.000000
Dhrs4 Ccnd2 Elavl4 Ank3 Klf12
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 0.000000 9.435728 9.435728 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 9.518087 0.000000 9.518087 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 10.924000 9.132330 0.000000 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 0.000000 10.000473 10.916737 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 10.278290 10.971420 9.585177 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 9.688617 10.094061 10.094061 9.688617 0.000000
Slc17a6 Lzts1 Map2 Lhx2 Neurod2
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 0.000000 0.000000 10.128836 10.534287
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 9.518087 10.211198 9.518087 9.518087
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 10.518544 9.825424 10.230871 9.825424
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 10.405923 0.000000 10.000473 11.099055
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.000000 10.971420 10.971420 10.278290
10X74_4_A_1:CTAGGATGCTTGCCx 8.995532 10.381733 8.995532 10.787188 10.604871
Sstr2 Eomes Vps37b Ier5 Cttnbp2
10X74_4_A_1:GTAACGTGCCATAGx 10.82196 0.000000 0.00000 10.128836 0.00000
10X73_3_A_1:ACGCCGGATCGTAGx 10.21120 0.000000 0.00000 9.518087 11.12747
10X74_4_A_1:GATTCGGAGGTGAGx 10.51854 9.825424 9.13233 0.000000 10.23087
10X74_4_A_1:CTGAGCCTGGTAAAx 10.69360 0.000000 0.00000 0.000000 0.00000
10X74_4_A_1:CCTAGAGAGTCACAx 10.27829 10.278290 10.27829 9.585177 11.19456
10X74_4_A_1:CTAGGATGCTTGCCx 10.94134 0.000000 10.60487 0.000000 10.94134
Tmem178 Plxna4 Aff3 Tiam2 Ppp2r2b
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 9.435728 9.435728 9.435728 10.82196
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 0.000000 10.616651 10.211198 10.21120
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 0.000000 0.000000 10.230871 9.13233
10X74_4_A_1:CTGAGCCTGGTAAAx 10.405923 0.000000 0.000000 9.307371 10.91674
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.000000 0.000000 0.000000 10.27829
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 0.000000 0.000000 10.094061 10.09406
Zeb2 Gria2 Nbea Ptprd Sox5
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 10.534287 0.000000 9.435728 10.821963
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 10.211198 9.518087 0.000000 9.518087
10X74_4_A_1:GATTCGGAGGTGAGx 9.825424 9.132330 0.000000 0.000000 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 10.405923 0.000000 10.405923 10.000473
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 9.585177 0.000000 0.000000 10.278290
10X74_4_A_1:CTAGGATGCTTGCCx 10.381733 0.000000 0.000000 0.000000 0.000000
Mpped2 Bcl11b Epha5 Nol4 Grik2
10X74_4_A_1:GTAACGTGCCATAGx 10.534287 9.435728 10.53429 0.00000 9.435728
10X73_3_A_1:ACGCCGGATCGTAGx 10.211198 0.000000 10.61665 10.21120 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 10.924000 0.000000 10.23087 0.00000 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 9.307371 0.000000 10.00047 10.91674 0.000000
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.000000 10.68374 10.97142 0.000000
10X74_4_A_1:CTAGGATGCTTGCCx 9.688617 0.000000 10.38173 0.00000 8.995532
Mir124.2hg Pcp4 Ccser1 Kcnb2 Dscaml1
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 0.000000 11.63288 10.128836 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 11.127466 10.90433 0.000000 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 10.741682 0.00000 9.132330 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 0.000000 11.09906 0.000000 0.000000
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 10.971420 10.27829 0.000000 10.278290
10X74_4_A_1:CTAGGATGCTTGCCx 8.995532 9.688617 10.38173 8.995532 9.688617
Kcnh7 Pam Negr1 Ppm1e Myt1l
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 0.000000 9.435728 10.534287 11.515100
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 10.616651 10.616651 9.518087 10.211198
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 9.825424 10.518544 9.825424 0.000000
10X74_4_A_1:CTGAGCCTGGTAAAx 9.307371 9.307371 0.000000 0.000000 10.405923
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 10.683744 11.194560 0.000000 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 10.381733 8.995532 0.000000 9.688617 10.094061
Foxp2 Gpc6 Masp1 Serpini1 Pcdh7
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 10.12884 0.000000 9.435728 10.12884
10X73_3_A_1:ACGCCGGATCGTAGx 9.518087 10.21120 9.518087 0.000000 11.12747
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 0.00000 9.132330 0.000000 9.13233
10X74_4_A_1:CTGAGCCTGGTAAAx 10.405923 0.00000 10.693598 0.000000 0.00000
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.00000 0.000000 0.000000 10.97142
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 0.00000 9.688617 0.000000 10.09406
Rbfox1 Robo2 Fhod3 Rprm Gng3
10X74_4_A_1:GTAACGTGCCATAGx 10.821963 11.38157 9.435728 0.000000 10.128836
10X73_3_A_1:ACGCCGGATCGTAGx 11.463934 11.71525 10.211198 0.000000 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 11.32946 9.132330 9.825424 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 11.099055 10.69360 0.000000 0.000000 0.000000
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 10.68374 0.000000 0.000000 0.000000
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 11.19265 9.688617 8.995532 8.995532
Uchl1 Tmem176b Nrn1 Snap25 Nrg1
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 9.435728 0.000000 9.435728 11.04510
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 10.211198 0.000000 0.000000 10.90433
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 9.132330 9.825424 0.000000 10.74168
10X74_4_A_1:CTGAGCCTGGTAAAx 9.307371 0.000000 10.916737 9.307371 11.38673
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 9.585177 9.585177 9.585177 10.97142
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 8.995532 9.688617 9.688617 11.19265
Cdh8 Trim17 Chga Rpl26 Rack1
10X74_4_A_1:GTAACGTGCCATAGx 10.12884 0.000000 0.000000 10.82196 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 11.30979 0.000000 9.518087 11.30979 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 10.74168 0.000000 0.000000 10.92400 10.230871
10X74_4_A_1:CTGAGCCTGGTAAAx 10.40592 9.307371 9.307371 10.91674 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 10.27829 9.585177 0.000000 10.68374 10.278290
10X74_4_A_1:CTAGGATGCTTGCCx 10.38173 8.995532 0.000000 11.29801 10.094061
Hsp90b1 Aldoa Tuba1b Sf3b6 Top1
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 9.435728 9.435728 0.00000 10.534287
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 9.518087 9.518087 0.00000 9.518087
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 9.132330 9.132330 9.13233 10.230871
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 9.307371 9.307371 0.00000 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.000000 0.000000 0.00000 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 8.995532 0.000000 10.09406 10.094061
Tpr Smc3 Rplp1 Rpl22 Rpl36a
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 0.000000 10.128836 0.000000 10.53429
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 0.000000 10.616651 0.000000 0.00000
10X74_4_A_1:GATTCGGAGGTGAGx 10.230871 0.000000 10.518544 9.132330 0.00000
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 0.000000 9.307371 10.000473 11.09906
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0.000000 10.683744 0.000000 10.68374
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 8.995532 10.094061 9.688617 10.09406
Rps21 Eif4a1 Erh Mdk Pde4d
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 9.435728 0.000000 0.00000 9.435728
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 10.211198 0.000000 10.61665 10.616651
10X74_4_A_1:GATTCGGAGGTGAGx 10.741682 9.825424 0.000000 11.07815 0.000000
10X74_4_A_1:CTGAGCCTGGTAAAx 9.307371 10.916737 0.000000 11.79219 0.000000
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 10.278290 0.000000 11.19456 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 10.381733 9.688617 9.688617 10.94134 10.941336
Pgrmc1 Ube2r2 Bri3 Rps28 Mbtd1
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 0.000000 0.000000 9.435728 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 9.518087 0.000000 9.518087 10.616651
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 0.000000 9.132330 9.132330 0.000000
10X74_4_A_1:CTGAGCCTGGTAAAx 9.307371 10.000473 9.307371 10.000473 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0.000000 9.585177 0.000000 10.683744
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 0.000000 9.688617 8.995532 9.688617
Rpl15 Macf1 Phf21a Oaz2 Gnb1
10X74_4_A_1:GTAACGTGCCATAGx 10.53429 9.435728 0.000000 10.12884 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 0.00000 9.518087 0.000000 0.00000 10.211198
10X74_4_A_1:GATTCGGAGGTGAGx 10.51854 0.000000 9.825424 9.13233 0.000000
10X74_4_A_1:CTGAGCCTGGTAAAx 10.40592 9.307371 0.000000 0.00000 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 10.68374 0.000000 0.000000 0.00000 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 10.60487 10.381733 8.995532 0.00000 0.000000
Zfp422 Rpl35a Cox8a Nt5dc2
10X74_4_A_1:GTAACGTGCCATAGx 10.128836 10.821963 10.128836 0.00000
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 9.518087 10.211198 0.00000
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 10.518544 9.132330 10.92400
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 10.000473 9.307371 10.00047
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.000000 0.000000 0.00000
10X74_4_A_1:CTAGGATGCTTGCCx 8.995532 9.688617 8.995532 10.09406
X4930402H24Rik Ttc28 Ckb Sh3bgrl
10X74_4_A_1:GTAACGTGCCATAGx 10.821963 10.128836 9.435728 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 9.518087 10.616651 0.000000 9.518087
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 10.741682 9.132330 0.000000
10X74_4_A_1:CTGAGCCTGGTAAAx 10.000473 11.253204 0.000000 0.000000
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 10.278290 0.000000 10.278290
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 9.688617 8.995532 9.688617
Top2b Ndufa10 C530008M17Rik Basp1 Gpatch8
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 0.00000 0.000000 10.534287 9.435728
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 0.00000 0.000000 10.616651 9.518087
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 10.23087 0.000000 10.518544 9.132330
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 10.00047 0.000000 9.307371 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.00000 9.585177 0.000000 0.000000
10X74_4_A_1:CTAGGATGCTTGCCx 8.995532 0.00000 10.094061 9.688617 8.995532
Map1b Ccdc88a Xist Ctnna2 Rnf7
10X74_4_A_1:GTAACGTGCCATAGx 10.534287 0.000000 0.00000 10.534287 9.435728
10X73_3_A_1:ACGCCGGATCGTAGx 9.518087 9.518087 10.61665 11.463934 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 0.000000 0.00000 9.825424 0.000000
10X74_4_A_1:CTGAGCCTGGTAAAx 10.000473 0.000000 10.69360 10.000473 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0.000000 11.78234 0.000000 0.000000
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 8.995532 10.38173 9.688617 9.688617
Fabp5 Usmg5 Phip Rbmx Zbtb20
10X74_4_A_1:GTAACGTGCCATAGx 0.000000 9.435728 9.435728 10.12884 9.435728
10X73_3_A_1:ACGCCGGATCGTAGx 9.518087 0.000000 0.000000 0.00000 10.211198
10X74_4_A_1:GATTCGGAGGTGAGx 9.825424 0.000000 9.132330 9.13233 10.518544
10X74_4_A_1:CTGAGCCTGGTAAAx 10.000473 9.307371 0.000000 0.00000 9.307371
10X74_4_A_1:CCTAGAGAGTCACAx 0.000000 0.000000 0.000000 0.00000 10.278290
10X74_4_A_1:CTAGGATGCTTGCCx 10.381733 8.995532 0.000000 0.00000 8.995532
Eif1b Lcor Nsg1 Evl Meg3
10X74_4_A_1:GTAACGTGCCATAGx 9.435728 0.000000 9.435728 9.435728 0.000000
10X73_3_A_1:ACGCCGGATCGTAGx 10.616651 9.518087 9.518087 9.518087 0.000000
10X74_4_A_1:GATTCGGAGGTGAGx 9.132330 9.825424 9.825424 0.000000 0.000000
10X74_4_A_1:CTGAGCCTGGTAAAx 10.693598 10.000473 10.693598 0.000000 10.000473
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0.000000 10.683744 0.000000 9.585177
10X74_4_A_1:CTAGGATGCTTGCCx 10.094061 0.000000 9.688617 8.995532 8.995532
Eif4a2 Rab10 Mllt11 COTAN_Cl.code
10X74_4_A_1:GTAACGTGCCATAGx 10.821963 0.000000 10.128836 0
10X73_3_A_1:ACGCCGGATCGTAGx 0.000000 9.518087 0.000000 0
10X74_4_A_1:GATTCGGAGGTGAGx 0.000000 10.230871 9.132330 0
10X74_4_A_1:CTGAGCCTGGTAAAx 0.000000 0.000000 10.405923 0
10X74_4_A_1:CCTAGAGAGTCACAx 9.585177 0.000000 9.585177 0
10X74_4_A_1:CTAGGATGCTTGCCx 0.000000 0.000000 10.381733 0# Fit the model
model <- glm( COTAN_Cl.code ~., data = train.data, family = binomial,control = list(maxit = 50))
# Summarize the model
summary(model)
Call:
glm(formula = COTAN_Cl.code ~ ., family = binomial, data = train.data,
control = list(maxit = 50))
Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -1.686e+15 5.418e+07 -31111735 <2e-16 ***
Abracl -4.715e+13 1.108e+06 -42568677 <2e-16 ***
Cdkn1c 1.198e+14 8.449e+05 141835339 <2e-16 ***
Cited2 -1.123e+14 1.158e+06 -96984394 <2e-16 ***
Mllt3 -1.480e+13 9.271e+05 -15963843 <2e-16 ***
Fam210b 5.776e+12 7.886e+05 7324089 <2e-16 ***
Mfap4 1.224e+13 1.105e+06 11082390 <2e-16 ***
Cxcl12 9.768e+12 8.228e+05 11871498 <2e-16 ***
Cdc42ep3 2.679e+13 8.102e+05 33062536 <2e-16 ***
Limch1 3.888e+13 7.629e+05 50956797 <2e-16 ***
Ier2 -1.941e+13 8.096e+05 -23978642 <2e-16 ***
Ptprk -1.101e+13 7.877e+05 -13974326 <2e-16 ***
Gm29260 -1.365e+14 8.513e+05 -160329184 <2e-16 ***
Sfrp1 -5.195e+13 7.665e+05 -67769176 <2e-16 ***
Pantr1 -6.604e+13 1.184e+06 -55777291 <2e-16 ***
Magi2 2.868e+13 7.718e+05 37162645 <2e-16 ***
Tmem108 2.640e+13 8.406e+05 31407609 <2e-16 ***
Ddr1 -7.295e+13 7.777e+05 -93801889 <2e-16 ***
Ezr -4.928e+12 8.165e+05 -6035217 <2e-16 ***
Sncaip -9.105e+13 7.615e+05 -119561130 <2e-16 ***
CT025619.1 7.755e+13 7.693e+05 100806643 <2e-16 ***
Fabp7 -7.246e+12 7.987e+05 -9071616 <2e-16 ***
Ctnnd2 4.181e+13 7.989e+05 52332613 <2e-16 ***
Ptn -6.993e+13 8.866e+05 -78870444 <2e-16 ***
Frmd4a -8.510e+12 8.139e+05 -10455059 <2e-16 ***
Sorbs2 8.519e+13 1.040e+06 81931351 <2e-16 ***
Hes6 4.145e+13 8.207e+05 50507058 <2e-16 ***
Dcc -7.059e+13 1.091e+06 -64717826 <2e-16 ***
Neurog2 1.400e+13 8.991e+05 15569920 <2e-16 ***
Tcf12 5.909e+13 7.580e+05 77960833 <2e-16 ***
Fam110a -1.887e+13 7.726e+05 -24430882 <2e-16 ***
Dhrs4 -8.288e+13 8.280e+05 -100100195 <2e-16 ***
Ccnd2 -6.734e+12 8.477e+05 -7943945 <2e-16 ***
Elavl4 7.681e+13 1.040e+06 73878040 <2e-16 ***
Ank3 3.085e+13 8.158e+05 37815294 <2e-16 ***
Klf12 2.096e+13 8.113e+05 25830911 <2e-16 ***
Slc17a6 4.704e+13 7.777e+05 60486563 <2e-16 ***
Lzts1 1.221e+13 7.629e+05 16006958 <2e-16 ***
Map2 -2.064e+13 1.049e+06 -19667725 <2e-16 ***
Lhx2 7.691e+13 1.073e+06 71697568 <2e-16 ***
Neurod2 1.553e+14 1.308e+06 118766112 <2e-16 ***
Sstr2 -1.201e+13 1.123e+06 -10693159 <2e-16 ***
Eomes 4.862e+13 8.998e+05 54037701 <2e-16 ***
Vps37b 5.989e+13 7.719e+05 77580428 <2e-16 ***
Ier5 -7.104e+13 7.933e+05 -89558067 <2e-16 ***
Cttnbp2 -6.238e+12 1.026e+06 -6082076 <2e-16 ***
Tmem178 4.639e+13 7.711e+05 60164738 <2e-16 ***
Plxna4 2.699e+13 8.247e+05 32726248 <2e-16 ***
Aff3 2.566e+13 8.494e+05 30215978 <2e-16 ***
Tiam2 3.867e+13 8.431e+05 45865723 <2e-16 ***
Ppp2r2b 7.367e+12 1.343e+06 5487241 <2e-16 ***
Zeb2 -8.699e+13 8.214e+05 -105905030 <2e-16 ***
Gria2 4.712e+12 8.115e+05 5807339 <2e-16 ***
Nbea -3.057e+13 7.864e+05 -38878840 <2e-16 ***
Ptprd -1.635e+13 8.432e+05 -19394693 <2e-16 ***
Sox5 7.997e+11 9.521e+05 839948 <2e-16 ***
Mpped2 -1.178e+13 9.316e+05 -12640992 <2e-16 ***
Bcl11b -8.699e+13 8.186e+05 -106261338 <2e-16 ***
Epha5 7.789e+13 1.438e+06 54158015 <2e-16 ***
Nol4 4.350e+13 8.773e+05 49578826 <2e-16 ***
Grik2 2.092e+13 7.828e+05 26717635 <2e-16 ***
Mir124.2hg -1.978e+13 8.306e+05 -23811113 <2e-16 ***
Pcp4 -7.976e+12 7.703e+05 -10354354 <2e-16 ***
Ccser1 7.204e+13 1.011e+06 71225614 <2e-16 ***
Kcnb2 -6.638e+13 7.809e+05 -85011877 <2e-16 ***
Dscaml1 -2.009e+13 8.061e+05 -24923932 <2e-16 ***
Kcnh7 -6.348e+13 8.216e+05 -77260888 <2e-16 ***
Pam -1.908e+13 9.566e+05 -19940070 <2e-16 ***
Negr1 -5.593e+13 8.125e+05 -68845519 <2e-16 ***
Ppm1e -5.188e+12 7.777e+05 -6670744 <2e-16 ***
Myt1l -1.775e+13 9.473e+05 -18739638 <2e-16 ***
Foxp2 1.982e+13 7.833e+05 25304938 <2e-16 ***
Gpc6 4.766e+12 7.945e+05 5999112 <2e-16 ***
Masp1 4.384e+13 7.665e+05 57188102 <2e-16 ***
Serpini1 2.141e+13 8.061e+05 26565876 <2e-16 ***
Pcdh7 4.264e+13 8.073e+05 52814591 <2e-16 ***
Rbfox1 -2.540e+12 8.031e+05 -3162771 <2e-16 ***
Robo2 -1.937e+13 1.420e+06 -13639796 <2e-16 ***
Fhod3 4.406e+13 7.980e+05 55209662 <2e-16 ***
Rprm 5.339e+13 7.835e+05 68145974 <2e-16 ***
Gng3 -2.037e+13 7.680e+05 -26522083 <2e-16 ***
Uchl1 7.239e+12 7.594e+05 9531581 <2e-16 ***
Tmem176b 3.088e+13 8.030e+05 38459013 <2e-16 ***
Nrn1 -5.860e+13 9.557e+05 -61323738 <2e-16 ***
Snap25 -2.782e+13 7.898e+05 -35221750 <2e-16 ***
Nrg1 -1.037e+14 1.511e+06 -68676296 <2e-16 ***
Cdh8 -6.110e+13 9.459e+05 -64591781 <2e-16 ***
Trim17 -4.393e+13 7.920e+05 -55467510 <2e-16 ***
Chga 3.352e+13 7.864e+05 42626343 <2e-16 ***
Rpl26 -1.541e+14 1.158e+06 -133038070 <2e-16 ***
Rack1 -1.958e+13 1.065e+06 -18388573 <2e-16 ***
Hsp90b1 1.180e+13 7.764e+05 15202488 <2e-16 ***
Aldoa 5.076e+13 7.895e+05 64297711 <2e-16 ***
Tuba1b -2.882e+13 7.673e+05 -37556112 <2e-16 ***
Sf3b6 -8.811e+12 7.754e+05 -11363341 <2e-16 ***
Top1 -7.822e+13 1.256e+06 -62293981 <2e-16 ***
Tpr 4.572e+13 8.111e+05 56364549 <2e-16 ***
Smc3 7.762e+12 7.864e+05 9869483 <2e-16 ***
Rplp1 4.751e+13 1.506e+06 31536103 <2e-16 ***
Rpl22 5.640e+13 9.818e+05 57449909 <2e-16 ***
Rpl36a 5.228e+13 1.286e+06 40661251 <2e-16 ***
Rps21 -6.595e+12 9.803e+05 -6728251 <2e-16 ***
Eif4a1 5.623e+13 1.107e+06 50779647 <2e-16 ***
Erh 5.627e+13 8.249e+05 68215509 <2e-16 ***
Mdk 1.403e+13 1.311e+06 10698211 <2e-16 ***
Pde4d 7.496e+13 8.054e+05 93072149 <2e-16 ***
Pgrmc1 2.521e+13 7.890e+05 31958231 <2e-16 ***
Ube2r2 2.252e+13 7.630e+05 29515130 <2e-16 ***
Bri3 -1.631e+13 7.624e+05 -21389061 <2e-16 ***
Rps28 5.956e+12 1.081e+06 5511164 <2e-16 ***
Mbtd1 -1.601e+13 8.020e+05 -19968438 <2e-16 ***
Rpl15 1.769e+13 1.056e+06 16743069 <2e-16 ***
Macf1 -2.359e+13 7.456e+05 -31643581 <2e-16 ***
Phf21a -2.270e+13 7.941e+05 -28584159 <2e-16 ***
Oaz2 -1.602e+13 7.671e+05 -20878553 <2e-16 ***
Gnb1 2.115e+13 7.786e+05 27161497 <2e-16 ***
Zfp422 -1.520e+13 7.937e+05 -19144492 <2e-16 ***
Rpl35a -7.101e+13 1.012e+06 -70148519 <2e-16 ***
Cox8a 8.371e+12 9.384e+05 8920618 <2e-16 ***
Nt5dc2 -3.363e+13 7.820e+05 -43000354 <2e-16 ***
X4930402H24Rik 1.789e+12 8.011e+05 2232956 <2e-16 ***
Ttc28 7.823e+12 9.395e+05 8326198 <2e-16 ***
Ckb -9.743e+13 8.412e+05 -115823225 <2e-16 ***
Sh3bgrl -2.330e+13 7.831e+05 -29748021 <2e-16 ***
Top2b 5.521e+13 7.850e+05 70331459 <2e-16 ***
Ndufa10 -7.000e+13 7.998e+05 -87527429 <2e-16 ***
C530008M17Rik -5.369e+13 8.134e+05 -66011040 <2e-16 ***
Basp1 1.162e+14 1.002e+06 115896331 <2e-16 ***
Gpatch8 1.888e+13 7.653e+05 24675271 <2e-16 ***
Map1b -2.937e+13 9.207e+05 -31901455 <2e-16 ***
Ccdc88a 3.041e+13 7.736e+05 39301701 <2e-16 ***
Xist 1.730e+13 7.861e+05 22010505 <2e-16 ***
Ctnna2 -4.172e+13 9.718e+05 -42928048 <2e-16 ***
Rnf7 -2.193e+13 8.067e+05 -27182858 <2e-16 ***
Fabp5 6.984e+13 9.924e+05 70373006 <2e-16 ***
Usmg5 5.091e+13 7.931e+05 64193211 <2e-16 ***
Phip 2.855e+13 7.823e+05 36499842 <2e-16 ***
Rbmx 1.080e+13 7.791e+05 13857287 <2e-16 ***
Zbtb20 -3.542e+13 1.022e+06 -34653461 <2e-16 ***
Eif1b -5.204e+13 1.019e+06 -51060909 <2e-16 ***
Lcor 1.071e+13 7.874e+05 13600823 <2e-16 ***
Nsg1 -6.878e+12 9.945e+05 -6915923 <2e-16 ***
Evl -3.260e+13 7.905e+05 -41243490 <2e-16 ***
Meg3 6.629e+13 8.077e+05 82077606 <2e-16 ***
Eif4a2 -3.223e+13 7.879e+05 -40902790 <2e-16 ***
Rab10 -3.146e+13 7.727e+05 -40719594 <2e-16 ***
Mllt11 1.398e+14 1.233e+06 113408602 <2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
(Dispersion parameter for binomial family taken to be 1)
Null deviance: 653.13 on 472 degrees of freedom
Residual deviance: 4757.76 on 326 degrees of freedom
AIC: 5051.8
Number of Fisher Scoring iterations: 46# Make predictions
probabilities <- model %>% predict(test.data, type = "response")
predicted.classes <- ifelse(probabilities > 0.5, "1", "0")
# Model accuracy
mean(predicted.classes == test.data$COTAN_Cl.code)[1] 0.6694915434 e13.5 e14.5
Sys.time()[1] "2024-04-19 12:11:58 CEST"sessionInfo()R version 4.3.2 (2023-10-31)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 20.04.6 LTS
Matrix products: default
BLAS: /usr/lib/x86_64-linux-gnu/openblas-pthread/libblas.so.3
LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/liblapack.so.3; LAPACK version 3.9.0
locale:
[1] LC_CTYPE=C.UTF-8 LC_NUMERIC=C LC_TIME=C.UTF-8
[4] LC_COLLATE=C.UTF-8 LC_MONETARY=C.UTF-8 LC_MESSAGES=C.UTF-8
[7] LC_PAPER=C.UTF-8 LC_NAME=C LC_ADDRESS=C
[10] LC_TELEPHONE=C LC_MEASUREMENT=C.UTF-8 LC_IDENTIFICATION=C
time zone: Europe/Rome
tzcode source: system (glibc)
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] torch_0.12.0 nnet_7.3-19 COTAN_2.5.0 caret_6.0-94
[5] lattice_0.22-5 lubridate_1.9.2 forcats_1.0.0 stringr_1.5.0
[9] dplyr_1.1.2 purrr_1.0.1 readr_2.1.4 tibble_3.2.1
[13] tidyverse_2.0.0 tidyr_1.3.0 parallelDist_0.2.6 data.table_1.15.0
[17] zeallot_0.1.0 ggplot2_3.5.0
loaded via a namespace (and not attached):
[1] RcppAnnoy_0.0.21 splines_4.3.2
[3] later_1.3.1 polyclip_1.10-4
[5] hardhat_1.3.0 pROC_1.18.4
[7] rpart_4.1.23 fastDummies_1.7.3
[9] lifecycle_1.0.3 doParallel_1.0.17
[11] processx_3.8.2 globals_0.16.2
[13] MASS_7.3-60 dendextend_1.17.1
[15] magrittr_2.0.3 plotly_4.10.2
[17] rmarkdown_2.24 yaml_2.3.7
[19] httpuv_1.6.11 Seurat_5.0.0
[21] sctransform_0.4.1 askpass_1.2.0
[23] spam_2.10-0 spatstat.sparse_3.0-2
[25] sp_2.1-1 reticulate_1.35.0
[27] cowplot_1.1.1 pbapply_1.7-2
[29] RColorBrewer_1.1-3 abind_1.4-5
[31] Rtsne_0.17 BiocGenerics_0.46.0
[33] coro_1.0.4 ipred_0.9-14
[35] circlize_0.4.15 lava_1.8.0
[37] IRanges_2.34.1 S4Vectors_0.38.1
[39] ggrepel_0.9.5 irlba_2.3.5.1
[41] spatstat.utils_3.0-3 listenv_0.9.0
[43] umap_0.2.10.0 goftest_1.2-3
[45] RSpectra_0.16-1 spatstat.random_3.2-1
[47] dqrng_0.3.0 fitdistrplus_1.1-11
[49] parallelly_1.37.1 DelayedMatrixStats_1.22.5
[51] leiden_0.4.3 codetools_0.2-19
[53] DelayedArray_0.26.7 tidyselect_1.2.0
[55] shape_1.4.6 farver_2.1.1
[57] viridis_0.6.4 ScaledMatrix_1.8.1
[59] spatstat.explore_3.2-1 matrixStats_1.2.0
[61] stats4_4.3.2 jsonlite_1.8.7
[63] GetoptLong_1.0.5 ellipsis_0.3.2
[65] progressr_0.14.0 ggridges_0.5.4
[67] survival_3.5-8 iterators_1.0.14
[69] foreach_1.5.2 tools_4.3.2
[71] ica_1.0-3 Rcpp_1.0.11
[73] glue_1.7.0 gridExtra_2.3
[75] prodlim_2023.08.28 xfun_0.39
[77] MatrixGenerics_1.12.3 ggthemes_5.1.0
[79] withr_3.0.0 fastmap_1.1.1
[81] fansi_1.0.4 openssl_2.1.0
[83] callr_3.7.3 digest_0.6.33
[85] rsvd_1.0.5 timechange_0.3.0
[87] R6_2.5.1 mime_0.12
[89] colorspace_2.1-0 scattermore_1.2
[91] tensor_1.5 spatstat.data_3.0-1
[93] utf8_1.2.3 generics_0.1.3
[95] recipes_1.0.8 class_7.3-22
[97] httr_1.4.6 htmlwidgets_1.6.2
[99] S4Arrays_1.2.0 uwot_0.1.16
[101] ModelMetrics_1.2.2.2 pkgconfig_2.0.3
[103] gtable_0.3.3 timeDate_4032.109
[105] ComplexHeatmap_2.16.0 lmtest_0.9-40
[107] htmltools_0.5.8 dotCall64_1.1-0
[109] clue_0.3-64 SeuratObject_5.0.0
[111] scales_1.3.0 png_0.1-8
[113] gower_1.0.1 knitr_1.43
[115] rstudioapi_0.15.0 tzdb_0.4.0
[117] reshape2_1.4.4 rjson_0.2.21
[119] nlme_3.1-163 zoo_1.8-12
[121] GlobalOptions_0.1.2 KernSmooth_2.23-22
[123] parallel_4.3.2 miniUI_0.1.1.1
[125] RcppZiggurat_0.1.6 pillar_1.9.0
[127] grid_4.3.2 vctrs_0.6.3
[129] RANN_2.6.1 promises_1.2.0.1
[131] BiocSingular_1.16.0 beachmat_2.16.0
[133] xtable_1.8-4 cluster_2.1.6
[135] evaluate_0.21 cli_3.6.1
[137] compiler_4.3.2 rlang_1.1.1
[139] crayon_1.5.2 future.apply_1.11.0
[141] labeling_0.4.2 ps_1.7.5
[143] plyr_1.8.8 stringi_1.8.1
[145] deldir_2.0-2 viridisLite_0.4.2
[147] BiocParallel_1.34.2 assertthat_0.2.1
[149] munsell_0.5.0 lazyeval_0.2.2
[151] spatstat.geom_3.2-4 PCAtools_2.14.0
[153] Matrix_1.6-3 RcppHNSW_0.6.0
[155] hms_1.1.3 patchwork_1.2.0
[157] bit64_4.0.5 sparseMatrixStats_1.12.2
[159] future_1.33.0 shiny_1.8.0
[161] ROCR_1.0-11 Rfast_2.1.0
[163] igraph_2.0.3 RcppParallel_5.1.7
[165] bit_4.0.5