In neurodata/r-mgc: Multiscale Graph Correlation

require(fmriutils)
require(graphstats)
require(mgc)
require(ggplot2)
require(latex2exp)
require(igraph)
require(stringr)
require(gridExtra)
require(scales)
require(data.table)
require(grid)
require(graphstats)

The data below can be downloaded and moved to appropriate folders as follows (note that the below section requires sudo access) from m2g.io.

Diffusion

nroi <- 70
dwi.dsets = c('BNU1', 'BNU3', 'HNU1', 'KKI2009', 'NKI1', 'NKIENH', 'MRN1313', 'Templeton114', 'Templeton255', 'SWU4')
dwi.atlas = 'desikan'
dwi.basepath = '/data/all_mr/dwi/edgelists'

graphobj = fmriu.io.collection.open_graphs(basepath = dwi.basepath, atlases = dwi.atlas, datasets = dwi.dsets,
                                           gname = 'graphs', fmt='edgelist', rtype = 'array', flatten = TRUE)
dwi.graphs = graphobj$graphs
dwi.datasets = graphobj$dataset
dwi.subjects = graphobj$subjects
dwi.sessions = graphobj$sessions

Discriminability

print("dMRI Discriminability")
dwi.results <- lapply(unique(dwi.datasets), function(dset) {
  ss <- dwi.datasets == dset
  graph.ss <- dwi.graphs[ss,]
  subs.ss <- dwi.subjects[ss]
  stat <- discr.stat(graph.ss, subs.ss)
  print(sprintf("%s dMRI: %.4f", dset, stat))
  return(data.frame(dataset=dset, modality='dMRI', discr=stat))
})
dwi.results <- do.call(rbind, dwi.results)

Functional MRI

nroi <- 70
fmri.dsets = c('BNU1', 'BNU2', 'BNU3', 'HNU1', 'IBATRT', 'IPCAS1', 'IPCAS2', 'IPCAS5', 'IPCAS6', 'IPCAS8', 'MRN1', 'NYU1', 'SWU1', 'SWU2', 'SWU3', 'SWU4', 'UWM', 'XHCUMS')
fmri.atlas = 'desikan-2mm'
fmri.basepath = '/data/all_mr/fmri/ranked/edgelists'

graphobj = fmriu.io.collection.open_graphs(basepath = fmri.basepath, atlases = fmri.atlas, datasets=fmri.dsets,
                                           fmt='edgelist', rtype = 'array', flatten=TRUE)
fmri.graphs = graphobj$graphs
fmri.datasets = graphobj$dataset
fmri.subjects = graphobj$subjects
fmri.sessions <- graphobj$sessions

Discriminability

print("fMRI Discriminability")
fmri.results <- lapply(unique(fmri.datasets), function(dset) {
  ss <- fmri.datasets == dset
  graph.ss <- fmri.graphs[ss,]
  subs.ss <- fmri.subjects[ss]
  stat <- discr.stat(graph.ss, subs.ss)
  print(sprintf("%s fMRI: %.4f", dset, stat))
  return(data.frame(dataset=dset, modality='fMRI', discr=stat))
})
fmri.results <- do.call(rbind, fmri.results)

Multimodal

ss.dsets <- fmri.dsets[which(fmri.dsets %in% dwi.dsets)]
fmri.graphs.ss <- fmri.graphs[fmri.datasets %in% ss.dsets,]
dwi.graphs.ss <- dwi.graphs[dwi.datasets %in% ss.dsets,]
dwi.graphs.ss <- t(apply(dwi.graphs.ss, c(1), rank))
fmri.datasets.ss <- fmri.datasets[fmri.datasets %in% ss.dsets]
dwi.datasets.ss <- dwi.datasets[dwi.datasets %in% ss.dsets]
fmri.subjects.ss <- fmri.subjects[fmri.datasets %in% ss.dsets]
dwi.subjects.ss <- dwi.subjects[dwi.datasets %in% ss.dsets]
fmri.sessions.ss <- fmri.sessions[fmri.datasets %in% ss.dsets]
dwi.sessions.ss <- dwi.sessions[dwi.datasets %in% ss.dsets]
fmri.data <- data.frame(subject=fmri.subjects.ss, session=fmri.sessions.ss, datasets=fmri.datasets.ss, data=fmri.graphs.ss)
dwi.data <- data.frame(subject=dwi.subjects.ss, session=dwi.sessions.ss, datasets=dwi.datasets.ss, data=dwi.graphs.ss)
mm.data <- merge(fmri.data, dwi.data, by=c("datasets", "subject", "session"), all=FALSE)

mm.results <- lapply(ss.dsets, function(dset) {
  dset.dat <- mm.data[mm.data$datasets == dset,]
  dset.dat.array <- as.matrix(dset.dat[, sapply(names(mm.data), function (x) grepl('data\\.', x))])
  stat <- discr.stat(dset.dat.array, dset.dat$subject)
  print(sprintf("%s MM: %.4f", dset, stat))
  return(data.frame(dataset=dset, modality='MM', discr=stat))
})
mm.results <- do.call(rbind, mm.results)

fmri.results.ss <- fmri.results[fmri.results$dataset %in% ss.dsets,]
dwi.results.ss <- dwi.results[dwi.results$dataset %in% ss.dsets,]
merged.results <- rbind(mm.results, fmri.results.ss, dwi.results.ss)
ggplot(merged.results, aes(x=dataset, y=discr, color=modality)) +
  geom_point(size=2) +
  xlab("Dataset") +
  ylab("Discriminability") +
  ggtitle("Investigating Impact of Multi-Modal Connectomics on Discriminability") +
  theme_bw()