In tangchao7498/DecodeR: Barcode demultiplexing of native RNAs nanopore sequencing

knitr::opts_chunk$set(echo = TRUE, 
                      warning = FALSE, 
                      message = FALSE, 
                      fig.align = 'center')
knitr::opts_knit$set(root.dir = "/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex")
knitr::opts_knit$set(base.dir = "/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex")

library(caret)
library(pbapply)
library(parallel)
library(multiROC)
library(doParallel)
library(data.table)

library(doParallel)
cl <- makePSOCKcluster(20)
registerDoParallel(cl)

load("./analysis/11.Classifiers/01.ModelingData/Barcode_4.RData")
rm(list = c("TrainingReads", "TestReads", "TestData")); gc()

set.seed(123)
TrainingData <- as.data.table(TrainingData)[, .SD[sort(sample(.N, 10000))], Class]

fitControl <- trainControl(method = "repeatedcv",
                           number = 10,
                           repeats = 10)

naive_bayes

set.seed(825)
naive_bayes <- train(Class ~ ., 
                     data = TrainingData,
                     preProc = c("center", "scale", "YeoJohnson", "nzv"),
                     method = "naive_bayes",
                     trControl = fitControl,
                     verbose = FALSE,
                     tuneLength = 10,
                     metric = "Accuracy",
                     allowParallel = TRUE)
saveRDS(naive_bayes, file = paste0("./analysis/12.AlgorithmComparison/Version1/01.Models/NB_B4.Rds"))

rf

set.seed(825)
rf <- train(Class ~ ., 
            data = TrainingData,
            preProc = c("center", "scale", "YeoJohnson", "nzv"),
            method = "rf",
            trControl = fitControl,
            verbose = FALSE,
            tuneLength = 10,
            metric = "Accuracy",
            allowParallel = TRUE)
saveRDS(rf, file = paste0("./analysis/12.AlgorithmComparison/Version1/01.Models/RF_B4.Rds"))

pcaNNet

set.seed(825)
pcaNNet <- train(Class ~ ., 
                 data = TrainingData,
                 preProc = c("center", "scale", "YeoJohnson", "nzv"),
                 method = "pcaNNet",
                 trControl = fitControl,
                 verbose = FALSE,
                 metric = "Accuracy",
                 allowParallel = TRUE)
saveRDS(pcaNNet, file = paste0("./analysis/12.AlgorithmComparison/Version1/01.Models/NNet_B4.Rds"))

knn

set.seed(825)
knn <- train(Class ~ ., 
             data = TrainingData, 
             method = "knn", 
             tuneLength = 6,
             trControl = fitControl)
saveRDS(knn, file = paste0("./analysis/12.AlgorithmComparison/Version1/01.Models/KNN_B4.Rds"))

CART

set.seed(825)
CART <- train(Class ~ ., 
              data = TrainingData, 
              method = "treebag", 
              trControl = fitControl)
saveRDS(CART, file = paste0("./analysis/12.AlgorithmComparison/Version1/01.Models/CART_B4.Rds"))

AdaBoost

set.seed(825)
AdaBoost <- train(Class ~ ., 
                  data = TrainingData, 
                  method = "AdaBoost.M1", 
                  trControl = fitControl)
saveRDS(AdaBoost, file = paste0("./analysis/12.AlgorithmComparison/Version1/01.Models/AdaBoost_B4.Rds"))

stopCluster(cl)

tangchao7498/DecodeR documentation built on May 27, 2023, 7:21 p.m.