Added:
- Niche cluster similarity visualization
Changed:
- Make niche cluster visualization no longer dependent on NT score.
Added:
- Node size legend in niche cluster connectivities visualization
Changed:
- Make node color legend optional in niche cluster connectivities visualization
Fixed:
- Wrong processing for sample name using int
Changed:
- Make options and modules' name consistent
- NN/niche network
- GNN/graph neural network
- NT/niche trajectory
- Rename
createDataSet
toONTraC_NN
- Rename
NicheTrajectory
toONTraC_NT
- Rename
ONTraC_GP
toONTraC_GT
- Refine the optparser structures
- Extract preprocessing modules
Added:
- Niche cluster connectivity visualization for each sample
- Node colorbar for cluster connectivity visualization
scale factor
controlling the size of spatial-based plotsN_GCN_LAYERS
controlling the number of GCN layers in the model
Changed:
- Make
log
file optional forONTraC_analysis
Fixed:
- Issues caused
ONTraC_GNN
still need dataset input
Added:
ONTraC_GNN
for running GNN only- GNN parameters group validation
- More functions descriptions under GNN part
- Transparent edges and colorbar for niche cluster connectivities visualization
TSP
method for Niche Trajectory construction
Changed:
- rename
GP
toONTraC_GP
Fixed:
- Loading dataset error in
NicheTrajectory
- Log printing error on Windows
Added:
- Niche trajectory visualization suppression parameter
Added:
- Integrate mudole with other workflow
Added:
- GitHub conda building workflow
Added:
- test module
- GitHub pytest workflow
Added:
citation
info
Added:
N_LOCAL
parameter- Parameters validation for
niche_net_constr
Changed:
- Support Python 3.10 and 3.12
- Refined Package structures
- Change the name of
NTScore
toNicheTrajectory
- Skip violinplot about cell type density along NT score in case of more than 100 cell types input
Changed:
- Make the default value of beta consistent with our paper
- Update tutorials
Fixed:
- Input table handling: support int format for sample
Added:
- Cell type composition visualization suppression parameter
- Niche cluster loadings visualization suppression parameter
Changed:
- Instruction for
analysis
installation - Make log file and train loss in
analysis
part optional - Get
edge_index
directly inniche_net_constr
module - Make losses name consistent with the paper
Fixed:
- Incorrect version display
Added:
- Citation information
- New parameter
sample
inanalysis
module for plotting by sample - Multiple cell type check in input data
- More detailed log output
Changed:
- Flexible figure size for some visualization
- IO options validation and output directories creation logic
- Device validate and use logic
- dataset load logic
Fixed:
- Fixed cell type composition read in analysis
- SpatialOmicsDataset using old saved data
Removed:
- Moved example dataset to Zenodo
Added:
- Added
ONTraC_analysis
andanalysis
module - Added
niche cluster
part inGP.py
- Added version information at the top of the output
Changed:
- Update package structures
- Update imports in
createDataSet.py
,GP.py
, andNTScore.py
- Update package buildup settings in
pyproject.toml
andMANIFEST.in
Fixed:
- Fixed bug in
createDataSet.py
Added:
- Added
installation
tutorial - Added
reproducible codes
for paper
Changed:
- Refactored the
pyproject.toml
according to setuptools - Rename losses name to make them consistent with paper
Removed:
- Removed Unused imports and files
- Removed
setup.py
Added:
- Added
simulation
data and tutorial - Added
niche cluster
information output - Added
niche cluster
tutorial - Added duplicate
Cell_ID
handle
Changed:
- Make this repository public
Fixed:
- Fixed errors when there is only 1 sample for
post-analysis
tutorial
Added
- Added
post-analysis
tutorial
Changed:
- Updated dependent packages information
- Updated installation tutorial
- Updated
stereo-seq
example
Added
- Added package description
- Added
.gitignore
to remove unnecessary files - Added
ONTraC
for running all steps together - Added
NTScore
for generating NTScore from GNN output - Added pip installation support
Changed:
- New environment constructions
- Running process
- Uniform parameters control
- Output directories
createDataSet
generate cell type composition and GNN input from raw input- Input data format