code/scripts/pannuke_cv_preprocess.sh

#!/bin/bash
# Preprocess pannuke fold02 (d11) + fold03 (d12) for BOTH nnU-Net and U-Mamba
# (separate preprocessed dirs, separate envs), CPU-only, thread-capped. Copies the
# fixed splits into each preprocessed dataset.
set -u
cd /home/wzhang/LSC/Code/NPJ
source /opt/anaconda3/etc/profile.d/conda.sh
RAW=/home/wzhang/LSC/Code/NPJ/nnunet_workspace/raw
export OMP_NUM_THREADS=4 MKL_NUM_THREADS=4 OPENBLAS_NUM_THREADS=4

( conda activate nnunet
  export nnUNet_raw=$RAW
  export nnUNet_preprocessed=/home/wzhang/LSC/Code/NPJ/nnunet_workspace/preprocessed_nnunet
  export nnUNet_results=/home/wzhang/LSC/Code/NPJ/nnunet_workspace/results_nnunet
  for id in 11 12; do
    nnUNetv2_plan_and_preprocess -d "$id" -c 2d -np 12 > "nnunet_workspace/pp_nnunet_d${id}.log" 2>&1
    name=$(basename "$(ls -d $RAW/Dataset0${id}_*)")
    cp "$RAW/$name/splits_final.json" "$nnUNet_preprocessed/$name/splits_final.json"
    echo "nnunet d$id preprocessed ($name)"
  done ) &

( conda activate umamba
  export nnUNet_raw=$RAW
  export nnUNet_preprocessed=/home/wzhang/LSC/Code/NPJ/nnunet_workspace/preprocessed_umamba
  export nnUNet_results=/home/wzhang/LSC/Code/NPJ/nnunet_workspace/results_umamba
  for id in 11 12; do
    nnUNetv2_plan_and_preprocess -d "$id" -c 2d -np 12 > "nnunet_workspace/pp_umamba_d${id}.log" 2>&1
    name=$(basename "$(ls -d $RAW/Dataset0${id}_*)")
    cp "$RAW/$name/splits_final.json" "$nnUNet_preprocessed/$name/splits_final.json"
    echo "umamba d$id preprocessed ($name)"
  done ) &

wait
echo PANNUKE_PP_DONE