#### Importing Data ####__________________________________________________________________________  
setwd("~/Desktop/Masters/2. Morphometrics/Morphometrics Data Collection/GMM Analysis/Datasets:Analyses/")
rearranged_gecko<-read.csv("SymmetricShape2020.csv", h=T)

  # Flip Z-Axis to Mirror Lefts on Rights
rawdata <- arrayspecs(rearranged_gecko[,c(5:ncol(rearranged_gecko))], p=55, k=3)
dim(rawdata) # [1] 55  3 30, p k n
rawdata[,3,which(rearranged_gecko$Side=="Left")] <- rawdata[,3,which(rearranged_gecko$Side=="Left")] * rep(-1,dim(rawdata)[1])
gecko_matrix<-rawdata

  # Defining Curves - Geomorph
define.sliders(gecko_matrix, 40, write.file=T)
geckosliders <- rbind(define.sliders(16:25), define.sliders(26:35), define.sliders(36:45), define.sliders(46:55))

#### Assymetry ####__________________________________________________________________________ 
# Assessing Asymmetry in Left-Right Maxillae Shape (+ Measurement Error)
ind <- rearranged_gecko$Specimen
side <- rearranged_gecko$Side
rep <- rearranged_gecko$Replicate
max.symmetry<-bilat.symmetry(fixed_gecko.gpa$coords, ind = ind, side = side, replicate = rep, object.sym = FALSE, iter = 9999)
max.symmetry$symm.shape 
max.symmetry$size.anova

#### PC Analysis ####__________________________________________________________________________  
  # Principal Component Analysis (PCA Plot)    
fixed_gecko.gpa <- gpagen(rawdata, curves=geckosliders)
plot(fixed_gecko.gpa)
gecko.PCA<-plotTangentSpace(fixed_gecko.gpa$coords, axis1 = 1, axis2 = 2, label=TRUE, groups = palette()[rearranged_gecko$Species]) # Morpho

#### Canonical Variate Analysis ####__________________________________________________________________________ 
cva.1<-CVA(gecko.PCA$pc.scores[,1:19], groups = rearranged_gecko$Species, rounds = 10000)
plot(cva.1$CVscores, col=cva.1$groups, pch=as.numeric(cva.1$groups), typ="n", asp=1,xlab=paste("1st Canonical Axis", paste(round(cva.1$Var[1,2],1),"%")), ylab=paste("2nd Canonical Axis", paste(round(cva.1$Var[2,2],1),"%")))
text(cva.1$CVscores, as.character(cva.1$groups),col=as.numeric(cva.1$groups), cex=.7)

  # Mahalanobis distance probabilities
cva.1$Dist$probsMaha

  # Visualize a shape change from score -5 to 5 for CV1/CV2
proc<-procSym(fixed_gecko.gpa$coords)
cvvis5 <- 5*cvall$CVvis[,1]+cvall$Grandm
cvvisNeg5 <- -5*cvall$CVvis[,1]+cvall$Grandm
cvvis5 <- showPC(cvvis5,proc$PCs,proc$mshape)
cvvisNeg5 <- showPC(cvvisNeg5,proc$PCs,proc$mshape)
deformGrid3d(cvvis5,cvvisNeg5,ngrid = 0)
CVA1p<-cvvis5
CVA1n<-cvvisNeg5
cvvis5_2 <- 5*cvall$CVvis[,2]+cvall$Grandm
cvvisNeg5_2 <- -5*cvall$CVvis[,2]+cvall$Grandm
cvvis5_2 <- showPC(cvvis5_2,proc$PCs,proc$mshape)
cvvisNeg5_2 <- showPC(cvvisNeg5_2,proc$PCs,proc$mshape)
deformGrid3d(cvvis5_2,cvvisNeg5_2,ngrid = 0)
CVA2p<-cvvis5_2
CVA2n<-cvvisNeg5_2
  
  # Classifying Unknown Specimens in CVA   
write.csv(gecko.PCA$pc.scores, "Gecko_PC_Scores.csv")
ModernPC<-read.csv("Modern_PC_Scores.csv", h=T)         # Separate known and unknown PC scores
SubfossilPC<-read.csv("Subfossil_PC_Scores.csv", h=T)
PCID<-read.csv("Taxonomic_ID.csv")
CVAProb<-typprobClass(SubfossilPC, ModernPC, groups = PCID$Species,small = T, method = "wilson", cova = T, outlier = 0.2, cv =T)
CVAProb$probsCV
CVAProb$groupaffin
CVAProb$probs
  
#### Linear Discriminant Analysis (LDA) ####________________________________________________________________________
  
  # Run LDA Analysis
gecko.lda<-lda(ModernPC, grouping = PCID$Species,CV=F) #Performs LDA
modern.pred<-predict(gecko.lda, ModernPC) #Get Modern Probabilities
sub.pred<-predict(gecko.lda, SubfossilPC) #Performs Prediction
  
  # See LDA Results  
subfossil.posteriors<-sub.pred$posterior
subfossil.class<-sub.pred$class
subfossil.scores<-sub.pred$x
  
  # Visualise LDA  
write.csv(modern.pred$x, "Modern_LD_Score.csv")  
write.csv(sub.pred$x, "Subfossil_LD_Score.csv")   
LDScores<-read.csv("LD_Scores.csv", h=T)
plot(LDScores$LD2 ~ LDScores$LD1, col=as.numeric(LDScores$Species))
text(LDScores$LD2 ~ LDScores$LD1, labels=LDScores$Specimen, data=LDScores, cex=0.3, font=0.5)
  
  # 95% Confidence Ellipse
dataEllipse(LDScores$LD1, LDScores$LD2, groups = LDScores$Species, levels=c(0.95))
    
  # Calculating Mean Shapes for Genera  
Dactylocnemis <- mshape(fixed_gecko.gpa$coords[,,1:9])
Hoplodactylus <- mshape(fixed_gecko.gpa$coords[,,10:16])
Mokopirirakau <- mshape(fixed_gecko.gpa$coords[,,17:21])
Naultinus <- mshape(fixed_gecko.gpa$coords[,,22:30])
Woodworthia <- mshape(fixed_gecko.gpa$coords[,,43:55])    

#### Procrustes ANOVA Analyses ####__________________________________________________________  

  # Multivariate Regression for Allometry (test for slope differences between groups)
MVAllom.1<-procD.lm(gecko.PCA$pc.scores[,1:19] ~ rearranged_gecko$Csize  * rearranged_gecko$Species, data = fixed_gecko.gpa, iter=9999)    
summary(MVAllom.1)
MVAllom.1.pair<-pairwise(MVAllom.1, groups = rearranged_gecko$Species)
summary(MVAllom.1.pair, confidence = 0.95)
    
  # ANOVA + Tukey-Test for Size Differences
bartlett.test(Csize ~ Species, data = rearranged_gecko)
csizeANOVA<-aov(Csize ~ Species, data = rearranged_gecko)
summary(csizeANOVA)
TukeyHSD(csizeANOVA)

  # Barplot_CSize_Means/SE    
means <- tapply(rearranged_gecko$Csize, rearranged_gecko$Species, FUN = mean, na.rm = T)
sds <- tapply(rearranged_gecko$Csize, rearranged_gecko$Species, FUN = sd, na.rm = T)
ns <- tapply(rearranged_gecko$Csize, rearranged_gecko$Species, FUN = length)
ses<-sds/sqrt(ns)
mp<-barplot(means, ylim=c(0,2100))
box(bty = "l")
segments(mp, means - 2*ses, mp, means + 2*ses, lwd=1.5)
segments(mp - 0.04, means - 2*ses, mp + 0.04, means - 2*ses, lwd=1.5)
segments(mp - 0.04, means + 2*ses, mp + 0.04, means + 2*ses, lwd=1.5)
stripchart(Csize ~ Species, data = rearranged_gecko,  vertical = T, add = T, pch = 21, method = "jitter")

#### Assessing Phylogenetic Signal ####______________________________________________________
  # Creating a Tree File
gecko.tree<-"geckos(((H._duvaucelii_S:0.37, H._duvaucelii:0.26):1.27, ((W._maculata:1.08, W._Otago_large:1.11):0.25, W._chrysosireticus:1.25):0.22):0.37, (((M._granulatus:0.65, M._granulatus_southern_NI:0.66):0.14, ((D._pacificus:0.68, D._mokohinaus:0.67):0.06, D._three_kings:0.62):0.29):0.13, ((N._elegans:0.19, N._punctatus:0.18):0.09, N._stellatus:0.27):0.86):0.25);"
cat(gecko.tree, file = "gecko.tree")
tree.gecko <- read.tree("gecko.tree")
plot(tree.gecko)
    
  # Creating Species Coordinate Means
new.coords<-coords.subset(fixed_gecko.gpa$coords, group = rearranged_gecko$Species)
species <- lapply(new.coords, mshape) 
species<-array(unlist(species), c(55, 3, length(species)))
dimnames(species)[[3]] <- levels(rearranged_gecko$Species)
    
  # Testing Significance 
PS.shape<-physignal(species, tree.gecko, iter=999)
summary(PS.shape)
plotGMPhyloMorphoSpace(tree.gecko, species)
  
#### Creating Warps of PC max ####________________________________________________________________________________
ref <- mshape(fixed_gecko.gpa$coords)
mesh <- read.ply("S.33703.3_New.ply")
mesh$material$color <- "moccasin" ; open3d(); shade3d(mesh)
mesh.coords <- read.morphologika("S.33703.3.txt")
plotspec(mesh,mesh.coords[,,1]/1000, centered = F)
refmesh <- warpRefMesh(mesh, mesh.coords[,,1]/1000, fixed_gecko.gpa$coords[,,42], centered=FALSE)
    
PCs <- c("mesh")
  attach(gecko.PCA$pc.shapes) # so we can call each PC shape from this list in the environment
    for(i in PCs){
      tmp <- plotRefToTarget(ref, get(i), method="surface", mesh = refmesh, mag=1)
      assign(paste("m",i,sep=""), tmp)
      writePLY(paste("~/Desktop/",i,".ply",sep=""),withColors=T)
      options(warn = 0)
    }
    
    FOV = 0 # sets parallel projection, rather than orthogonal
    open3d(); view3d(fov=0);shade3d(refmesh) # open a mesh on rgl
    # set by hand in rgl window, adjust specimen into dorsal positon then:
    usrMat.dorsal <- par3d()$userMatrix # save position matrix
    write.csv(usrMat.dorsal, "usrMat.dorsal.csv",row.names=F) # write to WD for later
    usrMat.lateral <- par3d()$userMatrix # again, adjust specimen into lateral positon then save
    write.csv(usrMat.lateral, "usrMat.lateral.csv",row.names=F)
    usrMat.posterior <- par3d()$userMatrix
    write.csv(usrMat.posterior, "usrMat.posterior.csv",row.names=F)
    
    usrMat.dorsal <- as.matrix(read.csv("usrMat.dorsal.csv", header = T))
    usrMat.lateral <- as.matrix(read.csv("usrMat.lateral.csv", header = T))
    usrMat.posterior <- as.matrix(read.csv("usrMat.posterior.csv", header = T))
    
    open3d(FOV=0, userMatrix = usrMat.dorsal) # for a dorsal view
    shade3d(mPC1min) # to see PC1min shape in dorsal view.
    open3d(FOV=0, userMatrix = usrMat.lateral) # for a lateral view
    open3d(FOV=0, userMatrix = usrMat.posterior) # for a posterior view
    
    views <- c("dorsal", "lateral", "posterior")
    PCs <- c("ref")
    for(i in PCs){
      for(j in views){
        open3d(FOV=0, userMatrix=as.name(paste("usrMat", j, sep=".")), windowRect=c(0,0,600,600))
        shade3d(get(paste("m",i,sep="")), alpha=1, lit = T, shininess = 128.0, specular=49)
        options(warn = 0)
        rgl.snapshot(paste("~/Desktop/MSc Chapters/3. Morphometrics/GMM Analysis/Datasets:Analyses/3DModels/Figures/","Mesh_",i,"_",j,".png", sep=""), fmt="png", top=TRUE)
        rgl.close()
      }}