FICT-SAMPLE: A Jupyter Notebook repository from jleechung

Sample code for FICT-FISH Iterative Cell Typing

Installation

git clone https://github.com/haotianteng/FICT-SAMPLE.git
cd FICT-SAMPLE
git clone https://github.com/haotianteng/FICT
git clone https://github.com/haotianteng/GECT
export PYTHONPATH="$(pwd)/FICT/:$(pwd)/GECT/:$PYTHONPATH"

Install dependency

conda activate YOUR_ENVIRONMENT
pip install -r requirements.txt

Preparing datasets

osmFISH: datasets/osmFISH_SScortex_mouse_all_cells.loom
seqFISH: datasets/seqFISH/{fcortex.coordinates, fcortex.expression, fcortex.genes}
MERFISH: datasets/aau5324_Moffitt_Table-S7.xlsx
*Download full MERFISH data matrix Moffitt_and_Bambah-Mukku_et_al_merfish_all_cells.csv from here. Unzip it to the datasets folder.

python prepare_osmFISH.py
python prepare_seqFISH.py
python prepare_MERFISH.py

Simulation

REPEAT=1
THREADS=1
python sim_from_real.py --output simulation
python dummy_train.py --prefix simulation/addictive -n $REPEAT --n_process $THREADS
python dummy_train.py --prefix simulation/exclusive -n $REPEAT --n_process $THREADS
python dummy_train.py --prefix simulation/stripe -n $REPEAT --n_process $THREADS
python dummy_train.py --prefix simulation/real -n $REPEAT --n_process $THREADS

Cross Validation

EMBEDDING_SIZE=20
python GECT/gect/gect_train_embedding.py -i Benchmark/MERFISH/data/1 -o Benchmark/MERFISH/ -m embedding --embedding-size $EMBEDDING_SIZE -b 200 --epoches 10 -t 4e-3
DATA=Benchmark/MERFISH/data
python FICT/cross_validation.py -i $DATA/1,$DATA/2,$DATA/3 -o Benchmark/MERFISH/FICT_CV/ --renew_round 40 --n_class 7 -d $EMBEDDING_SIZE --spatio_factor 0.1 --mode multi --reduced_method Embedding --embedding_file Benchmark/MERFISH/embedding

Cell typing

DATA=Benchmark/MERFISH/data/
python FICT/fict/run.py -i $DATA/0 -o Benchmark/Merfish/FICT_all/ --n_class 7

For the parameters of the cell typing, have a look at the run.py script, and also use help function