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")

library(PorexploreR)
library(data.table)
library(rhdf5)
library(ggplot2)
library(parallel)
library(changepoint)
library(caret)
library(pbapply)
library(GenomicAlignments)
library(ggbeeswarm)

readFast5 <- function(file, NT) {
  reads <- rhdf5::h5ls(file = file, recursive = FALSE)$name
  parallel::mclapply(X = reads, FUN = function(read) {
    # h5closeAll()
    h5_signals <- tryCatch(rhdf5::h5read(file = file, name = paste0(read, "/Raw"), read.attributes = TRUE), error = function(e) NA)
    if(is.na(h5_signals)) return(NA)
    signal_meta <- rhdf5::h5read(file, paste0(read, "/channel_id"), read.attributes = TRUE)

    range <- attr(signal_meta, "range")

    digitisation <- attr(signal_meta, "digitisation")
    scaling <- range/digitisation
    offset <- attr(signal_meta, "offset")
    sampling_rate <- attr(signal_meta, "sampling_rate")

    new("Squiggle", raw_signal = as.integer(h5_signals[[1]]), 
        range = as.numeric(range), digitisation = as.integer(digitisation), 
        offset = as.integer(offset), sampling_rate = as.integer(sampling_rate), 
        scaling = as.numeric(scaling))
  }, mc.cores = NT) -> res
  names(res) <- reads
  return(res)
}

normalize_signal <- function(sig) {
  med = median(sig)
  mad = median(abs(sig - med))
  (sig - med) / max(0.01, (mad * 1.4826))
}

GetBarcode3 <- function(read, length = 20000, plot = TRUE) {
  raw_sig <- PorexploreR::signal(read)
  if(length(raw_sig) > length) raw_sig <- raw_sig[seq_len(length)]

  Mat <- data.table::data.table(x = seq_along(raw_sig), norm = normalize_signal(raw_sig))
  Mat[, dema := smoother::smth(norm, method = "dema", n = 20)]

  cp2 <- Mat[!is.na(dema), suppressWarnings(changepoint::cpt.meanvar(dema, class = FALSE))[[1]]] + Mat[!is.na(dema), min(x)]

  ansmean <- tryCatch(suppressWarnings(changepoint::cpt.meanvar(Mat[cp2:nrow(Mat), dema], penalty = "Asymptotic", pen.value = 1e-10, method = "PELT")), error = function(e) NA)
  if(is.na(ansmean)) return(NULL)

  whichBin <- which.max(diff(c(0, head(ansmean@cpts, 4))))
  polyA_Pos <- ifelse(whichBin == 1, cp2, ansmean@cpts[whichBin - 1] + cp2)
  polyA_Sig <- ansmean@param.est$mean[whichBin]

  if(polyA_Pos < 2500) {
    BCSs <- NULL
  } else {
    if(mean(Mat[round(polyA_Pos*0.75):polyA_Pos, dema] < polyA_Sig) > 0.9) {
      # Mat[1:polyA_Pos,  smth := smoother::smth(norm, method = "dema", n = round(polyA_Pos/400), v = 1)]
      # BCSs <- Mat[!is.na(smth), smth]
      BCSs <- Mat[1:polyA_Pos, norm]
      if(plot) {
        Mat[1:polyA_Pos,  smth := smoother::smth(norm, method = "dema", n = round(polyA_Pos/400), v = 1)]
        Mat[, plot(x, norm, type = "s")]
        Mat[, lines(x, smth, type = "s", col = 2)]
        abline(v = polyA_Pos, col = 3, lty = 2, lwd = 3)
      }
    } else {
      if(mean(Mat[round(cp2*0.75):cp2, dema] < polyA_Sig) > 0.9) {
        BCSs <- Mat[1:cp2, norm]
        if(plot) {
          Mat[1:polyA_Pos,  smth := smoother::smth(norm, method = "dema", n = round(polyA_Pos/400), v = 1)]
          Mat[, plot(x, norm, type = "s")]
          Mat[1:cp2, lines(x, smth, type = "s", col = 2)]
          abline(v = cp2, col = 3, lty = 2, lwd = 3)
        }
      } else {
        BCSs <- NULL
      }
    }
  }

  if(!is.null(BCSs)) {
    if(length(BCSs) < 2100) {
      BCSs <- NULL
    }
  }

  return(BCSs)
}

MyChangePoint <- function(sig, MinLength = 10, ChangePoints = 68, StateStat = "Mean") {
  if(is.null(StateStat) | is.na(StateStat)) {
    stop("StateStat must be one of Mean or Median")
  }

  if(length(StateStat) != 1) {
    stop("StateStat must be one of Mean or Median")
  }

  if(!is.element(StateStat, c("Mean", "Median"))) {
    stop("StateStat must be one of Mean or Median")
  }

  cp0 <- suppressWarnings(changepoint::cpt.meanvar(data = sig, 
                                                   Q = ChangePoints, 
                                                   penalty = "Manual", 
                                                   method = "BinSeg", 
                                                   class = FALSE, 
                                                   minseglen = MinLength, 
                                                   param.estimates = FALSE, 
                                                   pen.value = 0.0001)) - 0.5
  bins <- cut(seq_along(sig), c(0, cp0, length(sig)), include.lowest = T, labels = FALSE)

  if(StateStat == "Mean") {
    bin_sig <- as.numeric(by(sig, bins, mean))
  } else {
    bin_sig <- as.numeric(by(sig, bins, median))
  }
  return(bin_sig)
}

files <- list.files("/mnt/raid61/Personal_data/songjunwei/DRS_RTA/batch4_f5_fq/20211022_1301_MN26652_FAQ89815_36b5e210/fast5_pass", "fast5$", recursive = TRUE, full.names = TRUE)

Barcode extraction

barcodeSigs <- lapply(files, function(file1) {
  print(file1)
  fast5_1 <- readFast5(file = file1, NT = 10)
  barcode <- mclapply(fast5_1, function(x) GetBarcode3(read = x, plot = F), mc.cores = 10)
  names(barcode) <- names(fast5_1)
  barcode[!mapply(is.null, barcode)]
})
barcodeSigs <- do.call(c, barcodeSigs)

Barcode segmentation

barcodeSigsBin <- pblapply(FUN = function(x) {
  MyChangePoint(sig = x, ChangePoints = 98, MinLength = 10, StateStat = "Mean")
}, barcodeSigs, cl = 4)

barcodeSigsBinMat <- as.data.frame(do.call(rbind, barcodeSigsBin))
colnames(barcodeSigsBinMat) <- paste0("BIN", sprintf("%03d", seq_len(ncol(barcodeSigsBinMat))))
saveRDS(barcodeSigsBinMat, "./analysis/08.m6A/01.Plasmodium/20211025/01.BarcodeDecomplex/barcodeSigsBinMat.Rds")

Label prediction

barcodeSigsBinMat <- readRDS("./analysis/08.m6A/01.Plasmodium/20211025/01.BarcodeDecomplex/barcodeSigsBinMat.Rds")
load(file = "./analysis/07.VirusClassification/03.Merge/01.Classifiers/20211008_6.RData")

Fit1

ROC_Tab <- data.frame(predict(Fit1, barcodeSigsBinMat, type = "prob"), 
                      pred = predict(Fit1, newdata = barcodeSigsBinMat))
saveRDS(ROC_Tab, "./analysis/08.m6A/01.Plasmodium/20211025/01.BarcodeDecomplex/BarcodesPrediction.Rds")

Preds <- readRDS("./analysis/08.m6A/01.Plasmodium/20211025/01.BarcodeDecomplex/BarcodesPrediction.Rds")
colnames(Preds) <- gsub("\\.", "-", colnames(Preds))
Preds <- as.data.table(Preds, keep.rownames = "qname")
Preds$PP <- apply(Preds[, grepl("RTA", colnames(Preds)), with = F], 1, max)

Preds[, .N, pred]
ggplot(Preds, aes(x = pred, y = PP)) + 
  geom_violin() + 
  scale_y_continuous(limits = c(0, 1)) + 
  theme_classic(base_size = 15)

Mat <- copy(Preds)
lapply(seq(0, 100, 5)/100, function(i) {
  ClassifiedReads <- Mat[, sum(PP >= i)]
  ReadsPercent <- Mat[, mean(PP >= i) * 100]
  data.frame(ClassifiedReads = ClassifiedReads, ClassifiedReadsPercent = ReadsPercent)
}) -> Cutoff_Select
Cutoff_Select <- as.data.table(do.call(rbind, Cutoff_Select))
Cutoff_Select$Cutoff <- seq(0, 100, 5)/100

ggplot(Cutoff_Select, aes(Cutoff, ClassifiedReadsPercent)) + 
  geom_line() + 
  labs(x = "PP", y = "Percentage of successful reads") + 
  theme_classic() + 
  theme(legend.position = "none", 
        axis.title = element_text(size = 16), 
        axis.text = element_text(size = 12)) + 
  geom_vline(xintercept = 0.3)

Preds[PP > 0.0, .N, pred][order(pred), ]
Preds[PP > 0.3, .N, pred][order(pred), ]
Preds[PP > 0.5, .N, pred][order(pred), ]

Preds[PP <= 0.3, pred := "Unclassified"]

library(microseq)
library(Biostrings)
fqfs <- list.files("/mnt/raid61/Personal_data/songjunwei/DRS_RTA/batch4_f5_fq/hac_fastq/pass", "fastq", full.names = TRUE)
fastq <- lapply(fqfs, microseq::readFastq)
fastq <- do.call(rbind, fastq)
id <- fastq$Header
fastq <- QualityScaledDNAStringSet(x = RNAStringSet(fastq$Sequence), 
                                   quality = PhredQuality(fastq$Quality))
names(fastq) <- id

setdiff(Preds[, levels(pred)], "Unclassified")
for(i in setdiff(Preds[, levels(pred)], "Unclassified")) {
  fi <- fastq[paste0("read_", mapply(function(x) x[1], strsplit(names(fastq), " "))) %in% Preds[pred == i, qname]]
  writeQualityScaledXStringSet(fi, paste0("./analysis/08.m6A/01.Plasmodium/20211025/02.SplitFastq/", i, ".fastq"))
}