add scgpt information

docs/devops.md

@@ -107,4 +107,18 @@ BiocManager::install("BiocNeighbors")
 install.packages('NMF')
 install.packages("circlize")
 devtools::install_github("jinworks/CellChat")
+```
+
+## Build environments
+
+### scGPT in FAS
+
+```
+conda create -p ./scgpt-2 python=3.9 pip ipykernel   -c conda-forge
+conda install ipywidgets -c conda-forge
+pip install torch==2.1.2
+conda install numpy
+pip install scgpt
+conda install wandb -c conda-forge
+python -m ipykernel install --prefix=/n/holylfs05/LABS/hsph_bioinfo/Lab/shared_resources/scgpt-2 --name 'dcgpt2' --display-name 'scgpt-2'
 ```

docs/environments.md

@@ -0,0 +1,21 @@
+# Environments availables
+
+## scGPT
+
+Only available at FAS computing resources.
+
+Please, reach out to platform to access for the first time to this env:
+
+- Start a python notebook on the [ondemand web-page](https://rcood.rc.fas.harvard.edu/pun/sys/dashboard/batch_connect/sessions). You need to be connected to the VPN to be connected. Use `gpu` partition and if you need more than one GPU, add these to the sbatch options `--gres=gpu:n`
+- Once connected (This step is tricky because you need to find the password to access the notebook)
+    - Initiate a terminal session
+    - `module load gcc/13.2.0-fasrc01`
+    - Only the first time, add the env to the notebook kernels:
+        - `python -m ipykernel install --prefix=/n/holylfs05/LABS/hsph_bioinfo/Lab/shared_resources/scgpt-2 --name 'dcgpt2' --display-name 'scgpt-2'`
+    - Then, start a notebook choosing as kernel `scgpt-2`
+- Validate everything works with these commands in the python notebook
+    - `import torch`
+    - `torch.cuda.is_available()` -> Should return `True`
+    - `torch.cuda.device_count()` -> Should return `
+    - `import scgpt`
+

mkdocs.yml

@@ -2,6 +2,7 @@ site_name: HCBC Platform
 nav:
   - Home: index.md
   - Pipelines: pipelines.md
+  - Tools environments: environments.md
   - Platform members: devops.md
 
 theme: