annotationRmn2D.R

###########################################################################################################################################
# ANNOTATION SPECTRE 2D MATRICE COMPLEXE BASEE SUR UNE SEQUENCE RMN                                                                       #
# matriceComplexe : data.frame liste couples ppm de la matrice a annoter                                                                  #
# BdDStandards : objet contenant la base de donnees des composes standards                                                                #
# nom_séquence : nom sequence 2D a utiliser pour annotation ("JRES","COSY","TOCSY","HMBC","HSQC")                                         #
# ppm1Tol : tolerance ppm axe abscisses                                                                                                   #
# ppm2Tol : tolerance ppm axe ordonnees                                                                                                   #
# nb_ligne_template : préciser le nombre total de ligne de la feuille de calcul à annoter                                                 #
###########################################################################################################################################
annotationRmn2D <- function(matriceComplexe, BdDStandards, nom_sequence, ppm1Tol=0.01, ppm2Tol=0.01, 
                            seuil=0, unicite="NO")
{
  ## Longueur de la peak-list de la matrice a annoter
  PeakListLength <- length(matriceComplexe[, 1])

  ## Nombre de metabolites inclus dans BdD de composes standards
  nbMetabolitesBdD <- length(BdDStandards)
  matrixAnnotation <- data.frame()
  allMetabolitesList <- data.frame()
  seuil_score <- seuil
  
  ## Boucle sur les metabolites inclus dans BdD
  for (i in 1:nbMetabolitesBdD)
  {
    ## Infos metabolite en cours
    iMetabolite <- BdDStandards[[i]]
    ppm1M <- iMetabolite[,1] 
    ppm2M <- iMetabolite[,2]
    nbPeakMetabolite <- length(ppm1M)
    MetaboliteName <- names(BdDStandards[i])
##    print(MetaboliteName)
    ## Initialisation
    k <- 0
    presenceScore <- 0
    annotatedPpmRef <- data.frame()
    annotatedPpmList <- data.frame()
    annotatedPeakLength <- 0
    metabolites <- data.frame()
    metabolitesList <- data.frame()
  
    ## Boucle sur les couples de pics de la matrice a annoter
    for (p in 1:PeakListLength)
    { 
      ppmAnnotationF1 <- as.numeric(matriceComplexe[p, 3])
      ppmAnnotationF2 <- as.numeric(matriceComplexe[p, 2])
      e <- simpleMessage("end of file")
      tryCatch({
        if (!is.na(ppmAnnotationF1))
        {
          matrixAnnotation <- unique.data.frame(rbind.data.frame(matrixAnnotation, matriceComplexe[p, ]))
        }
        # Recherche du couple de pics de la matrice la liste des couples du metabolite standard
        metaboliteIn <- (ppm1M >= (ppmAnnotationF2-ppm1Tol) & ppm1M <= (ppmAnnotationF2+ppm1Tol) & 
                     ppm2M >= (ppmAnnotationF1-ppm2Tol) & ppm2M <= (ppmAnnotationF1+ppm2Tol))
        WhichMetaboliteIn <- which(metaboliteIn)
        # Si au moins un couple de la matrice a annoter dans liste couples metabolite standard
        if (length(WhichMetaboliteIn) > 0)
        {
          for (a in 1:length(WhichMetaboliteIn))
          {
            annotatedPpmList <- data.frame(ppm1=ppm1M[WhichMetaboliteIn[a]], ppm2=ppm2M[WhichMetaboliteIn[a]], theoricalLength=nbPeakMetabolite)
            annotatedPpmRef <- rbind(annotatedPpmRef,annotatedPpmList)
          }
        }
      }, error=function(e){cat ("End of file \n");})
    }

    # Au - 1 couple de ppm de la matrice complexe annote
    if (nrow(annotatedPpmRef) >= 1)
    {
      ## Nombre couples annotes
      annotatedPeakLength <- nrow(annotatedPpmRef)
      
      ## Recherche doublons
      annotatedDoublons <- duplicated(annotatedPpmRef)
      if (sum(duplicated(annotatedPpmRef)) > 0)
      {
        annotatedPeakLength <- nrow(annotatedPpmRef) - sum(duplicated(annotatedPpmRef))
        annotatedPpmRef <- annotatedPpmRef[-duplicated(annotatedPpmRef), ]
      }
      presenceScore <- annotatedPeakLength/nbPeakMetabolite
    }
    
    ## Conservation metabolites dont score > seuil
    if (presenceScore > seuil_score)
    {
      metabolites <- data.frame(Metabolite=MetaboliteName, score=presenceScore)
      metabolitesList <- cbind.data.frame(annotatedPpmRef, metabolites) 
      allMetabolitesList <- rbind.data.frame(allMetabolitesList, metabolitesList)
    }
  }
  
  # Initialisation
  commonPpm <- data.frame()
  commonPpmList <- data.frame()
  metaboliteAdd <- data.frame()
  metaboliteAddList <- data.frame()
#  metabolite_ref <- data.frame()
  commonMetabolitesList <- data.frame()
  commonMetabolitesPpmList <- data.frame()
  commonMetabolitesPpmAllList1 <- data.frame()
  commonMetabolitesPpmAllList <- data.frame()
  listeTotale_2D_unicite <- allMetabolitesList[, 1:4]
  allMetabolitesList <- allMetabolitesList[, -3]
  metabolitesAllUnicite <- data.frame()
  
  ## Boucle sur tous couples annotes
  for (j in 1:length(allMetabolitesList$ppm1))
  {
    ## Boucle sur metabolites dans BdD composes standards
    for (i in 1:nbMetabolitesBdD)
    {
      ppmMetaboliteBdD <- BdDStandards[[i]]
      ppm1M <- ppmMetaboliteBdD[,1] 
      ppm2M <- ppmMetaboliteBdD[,2]
      # Nombre de couples metabolite
      nbPeakMetabolite <- length(ppm1M)
      MetaboliteName <- names(BdDStandards[i])

      metabolitesInAll <- (ppm1M >= (allMetabolitesList[j,1]-ppm1Tol) & ppm1M <= (allMetabolitesList[j,1]+ppm1Tol) & 
                            ppm2M >= (allMetabolitesList[j,2]-ppm2Tol) & ppm2M <= (allMetabolitesList[j,2]+ppm2Tol))
      WhichMetabolitesInAll <- which(metabolitesInAll)

      if (MetaboliteName != allMetabolitesList[j, 3] & length(WhichMetabolitesInAll) > 0)
      {
        metabolitesAllUnicite <- rbind.data.frame(metabolitesAllUnicite, listeTotale_2D_unicite[j,])
        commonPpm <- data.frame(ppm1=allMetabolitesList[j,1], ppm2=allMetabolitesList[j,2])
        commonPpmList <- rbind.data.frame(commonPpmList, commonPpm)
        commonPpmList <- unique(commonPpmList)
        metaboliteAdd <- data.frame(nom_metabolite=MetaboliteName)
        metaboliteAddList <- rbind.data.frame(metaboliteAddList, metaboliteAdd)
#        metabolite_ref <- data.frame(nom_metabolite=allMetabolitesList[j,3])
        commonMetabolitesList <- rbind.data.frame(data.frame(nom_metabolite=allMetabolitesList[j, 3]), metaboliteAddList)
        commonMetabolitesPpmList <- cbind.data.frame(commonPpm, commonMetabolitesList)
        commonMetabolitesPpmAllList1 <- rbind.data.frame(commonMetabolitesPpmAllList1, commonMetabolitesPpmList)
        commonMetabolitesPpmAllList1 <- unique.data.frame(commonMetabolitesPpmAllList1)
      }
    }
    commonMetabolitesPpmAllList <- rbind.data.frame(commonMetabolitesPpmAllList, commonMetabolitesPpmAllList1)
    commonMetabolitesPpmAllList <- unique.data.frame(commonMetabolitesPpmAllList)
    
    #initialisation des data.frame
    commonPpm <- data.frame()
    metaboliteAdd <- data.frame()
    metaboliteAddList <- data.frame()
    metabolite_ref <- data.frame()
    commonMetabolitesList <- data.frame()
    commonMetabolitesPpmList <- data.frame()
    commonMetabolitesPpmAllList1 <- data.frame()
  }

  unicityAllList <- listeTotale_2D_unicite
  if (nrow(listeTotale_2D_unicite)!=0 & nrow(metabolitesAllUnicite)!=0)
    unicityAllList <- setdiff(listeTotale_2D_unicite, metabolitesAllUnicite)

  unicitynbCouplesRectif <- data.frame()
  for (g in 1:nrow(unicityAllList))
  {
    metaboliteUnicity <- (unicityAllList$Metabolite == unicityAllList$Metabolite[g])
    WhichMetaboliteUnicity <- which(metaboliteUnicity)
    nb_occurence <- length(WhichMetaboliteUnicity)
    unicitynbCouplesRectif <- rbind.data.frame(unicitynbCouplesRectif, nb_occurence)
  }
  names(unicitynbCouplesRectif) <- "NbCouplesAnnotes"
  unicityAllList <- cbind.data.frame(unicityAllList, unicitynbCouplesRectif)
  
  unicityAllList <- cbind.data.frame(unicityAllList, score_unicite=unicityAllList$NbCouplesAnnotes/unicityAllList$theoricalLength)
  unicityAllList <- unicityAllList[, -3]
  unicityAllList <- unicityAllList[, -4]

##  unicityAllList <- filter(unicityAllList, unicityAllList$score_unicite > seuil_score)
  unicityAllList <- unicityAllList[unicityAllList$score_unicite > seuil_score,]

  listeTotale_metabo <- data.frame()
  if (nrow(commonPpmList) !=0)
  {
    for (o in 1:length(commonPpmList[, 1]))
    {
      tf6 <- (commonMetabolitesPpmAllList$ppm1 == commonPpmList[o,1] & commonMetabolitesPpmAllList$ppm2 == commonPpmList[o,2])
      w6 <- which(tf6) 
      
      for (s in 1:length(w6))
      {
        metaboliteAdd <- data.frame(nom_metabolite=commonMetabolitesPpmAllList[w6[s],3])
        commonMetabolitesList <- paste(commonMetabolitesList, metaboliteAdd[1,], sep = " ")
      }
      liste_metabo_ppm <- cbind.data.frame(ppm1=commonPpmList[o,1],ppm2=commonPpmList[o,2], commonMetabolitesList)
      listeTotale_metabo <- rbind.data.frame(listeTotale_metabo, liste_metabo_ppm)
      commonMetabolitesList <- data.frame()
    }
  }

  # Representation graphique
  if (nom_sequence == "HSQC" | nom_sequence == "HMBC")
  {
    atome <- "13C"
    indice_positif <- 1
    indice_negatif <- -10
  }else{
    atome <- "1H"
    indice_positif <- 0.5
    indice_negatif <- -0.5
  }
  
  matriceComplexe <- matrixAnnotation
  ppm1 <- as.numeric(matriceComplexe[,2])
  ppm2 <- as.numeric(matriceComplexe[,3])
  
  if (unicite == "NO")
  {
    listeTotale_2D_a_utiliser <- allMetabolitesList
    d1.ppm <- allMetabolitesList$ppm1 
    d2.ppm <- allMetabolitesList$ppm2
  }else{
    listeTotale_2D_a_utiliser <- unicityAllList
    d1.ppm <- listeTotale_2D_a_utiliser$ppm1 
    d2.ppm <- listeTotale_2D_a_utiliser$ppm2
  }

  if (nrow(listeTotale_2D_a_utiliser) > 0)
  {
    ## Taches de correlations
    # Matrice biologique + Annotations
    maxX <- max(round(max(as.numeric(matriceComplexe[,2])))+0.5, round(max(as.numeric(matriceComplexe[,2]))))
    maxY <- max(round(max(as.numeric(matriceComplexe[,3])))+indice_positif, round(max(as.numeric(matriceComplexe[,3]))))
    probability.score <- as.factor(round(listeTotale_2D_a_utiliser[,4],2))
    lgr <- length(unique(probability.score))
    sp <- ggplot(matriceComplexe, aes(x=ppm1, y=ppm2))
    sp <- sp + geom_point(size=2) + scale_x_reverse(breaks=seq(maxX, 0, -0.5)) + 
      scale_y_reverse(breaks=seq(maxY, 0, indice_negatif)) + 
      xlab("1H chemical shift (ppm)") + ylab(paste(atome, " chemical shift (ppm)")) + ggtitle(nom_sequence) +
      geom_text(data=listeTotale_2D_a_utiliser, aes(d1.ppm, d2.ppm, label=str_to_lower(substr(listeTotale_2D_a_utiliser[,3],1,3)), 
                                                    col=probability.score), 
                size=4, hjust=0, nudge_x=0.02, vjust=0, nudge_y=0.2) + scale_colour_manual(values=viridis(lgr))
##      scale_color_colormap('Annotation', discrete=T, reverse=T)
    print(sp)
  }
  
  # Liste des résultats (couples pmm / metabolite / score) + liste ppms metabolites communs
  if (unicite == "NO")
  {
    return(list(liste_resultat=allMetabolitesList, listing_ppm_commun=listeTotale_metabo))
  }else{
    return(list(liste_resultat_unicite=unicityAllList, listing_ppm_commun_affichage=listeTotale_metabo))
  }
}