Évaluer l'enrichissement

ChIP-seq avec Bioconductor en R

Peter Humburg

Statistician, Macquarie University

Extension des lectures

Charger les données :

reads <- readGAlignments(bam)
reads_gr <- granges(reads[[1]])

Obtenir la longueur moyenne des fragments :

frag_length <- fragmentlength(qc_report)["GSM1598218"]

Étendre les lectures et calculer la couverture :

reads_ext <- resize(reads_gr, width=frag_length)
cover_ext <- coverage(reads_ext)

Couverture des pics

Créer des blocs de 200 pb le long du génome.

bins <- tileGenome(seqinfo(reads), tilewidth=200, 
                   cut.last.tile.in.chrom=TRUE)

Trouver tous les blocs qui chevauchent des pics.

peak_bins_overlap <- findOverlaps(bins, peaks)
peak_bins <- bins[from(peak_bins_overlap), ]

Compter le nombre de lectures chevauchant chaque bloc de pic.

peak_bins$score <- countOverlaps(peak_bins, reads)

Fonction de couverture par blocs

count_bins <- function(reads, target, bins){
  # Find all bins overlapping peaks
  overlap <- from(findOverlaps(bins, target))
  target_bins <- bins[overlap, ]

  # Count the number of reads overlapping each peak bin
  target_bins$score <- countOverlaps(target_bins, reads)
  target_bins
}

Couverture des régions en liste noire

peak_bins <- count_bins(reads_ext, peaks, bins)
bl_bins <- count_bins(reads_ext, blacklist.hg19, bins)

Couverture de fond

Retirer tous les blocs déjà pris en compte.

bkg_bins <- subset(bins, !bins %in% peak_bins & !bins %in% bl_bins)

Compter les lectures chevauchant chaque bloc restant.

bkg_bins$score <- countOverlaps(bkg_bins, reads_ext)

Passons à la pratique !

ChIP-seq avec Bioconductor en R