README.md

In morphr/morphr: Shape analysis

Morphr (c) 2020

Morphr implements standard techniques for the representation of shapes, calculation of mean templates, statistical analysis (clustering, principal component analysis (PCA) and estimating deformations to mean shapes). It accepts case studies for medical, anatomical, and paleontological shape data. The package also implements convenient graphical user interface built using R shiny, allowing researchers with limited coding experience to conduct shape analysis case studies easily.

The idea behind Morphr is not only to provide tools for analysis of continuous shapes and curves, but also a set of integrated functionalities that can be used by morphologist with limited coding experience to conduct shape analysis case studies easily. The package also utilizes ideas and functions from elastic functional data analysis pacakge: fdasrvf.

Installation

• Install gfortran
• Install latest development version from GitHub (requires devtools package):

if (!require("devtools")) {
  install.packages("devtools")
}
devtools::install_github("morphr/morphr")

Examples

For demonstration, we present a case study for paleontological shapes. For a dataset of hadrosaurid bone specimens, we show results for building a mean template from a collection of shapes, summarize patterns of shape variation in PCA, and show hierarchical clusters obtained from pairwise distances between the collection. 

• Reconstruct Shape

path = 'Path for ucf or svg file list'
verify_shapes(path)

• Construct Mean shape

X = main_closed(path)
qarray = list()
for(i in 1:length(X)){
  qarray[[i]] = curve_to_q(X[[i]])
}
All_mean_shape = find_mean_shape(qarray)
saveRDS(All_mean_shape, file = "All_mean_shape.rds")
qmean = All_mean_shape[[1]]
qmean_new = project_curve(qmean)
pmean = q_to_curve(qmean_new)
plot_curve(pmean,'r',filename = 'Mean Shape')

• Calculate Geodesic Distance Heatmap

color_file_path = 'Path for colorcode and filename'
all_geodesic = geodesic_distance_all(qarray)
saveRDS(all_geodesic, file = "all_geodsic.rds")
geo_dist = all_geodesic[[6]]
output_result[lower.tri(output_result, diag=FALSE)] <- unlist(geo_dist)
   output_result <- t(output_result)
   for(i in 1:(length(X)-1)){
     for (j in (i+1):length(X)){
        output_result[j,i] = output_result[i,j]
      }
   }
rc_name <- c()
Taxoncolorcodes <- readr::read_csv(color_file_path,col_names = FALSE)
filenames <- Taxoncolorcodes$X1
for(i in 1:length(X)){
    rc_name <- c(rc_name,filenames[i])
}
rownames(output_result) <- rc_name
colnames(output_result) <- rc_name
plotly::plot_ly(x = rc_name, y = rc_name ,z = output_result, type = "heatmap")

• PCA Plot on Eigen Axis 1 and 2

alpha_t_array = All_mean_shape[[3]]
PCA_plot(alpha_t_array, qmean, qarray, color_file_path)

• PCA Variation Along Eigen Axis 1

plot_pca_variation(alpha_t_array, qmean, eigdir = 1,qarray, dt = 0)

• MDS Plot

mdsplot(alpha_t_array, geo_dist, X, color_file_path)

• Dendrogram

plot_dendrogram(geo_dist, color_file_path)

• Deformation Field

deformation_field_all(alpha_t_array, pmean, qmean,X)

morphr/morphr documentation built on Sept. 28, 2020, 5:39 p.m.