fix: use N=100 for FDR (matches paper), 100 trials, partial thresholds

- Original paper code used N=100 for get_thresh_FDR (not 5000)
- Add partial match FDR/FNR computation to SLURM scripts
- Update Apptainer base to PyTorch 2.4.0 (fixes glibc mismatch)
- Set 100 trials as default (stable estimates)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

apptainer.def

@@ -1,5 +1,5 @@
 Bootstrap: docker
-From: pytorch/pytorch:2.1.0-cuda12.1-cudnn8-runtime
+From: pytorch/pytorch:2.4.0-cuda12.1-cudnn9-runtime
 
 %labels
     Author Ron Boger <ronboger@berkeley.edu>

scripts/compute_fdr_table.py

@@ -94,8 +94,8 @@ def main():
     parser.add_argument(
         '--n-trials',
         type=int,
-        default=10,
-        help='Number of calibration trials (default: 10)'
+        default=100,
+        help='Number of calibration trials (default: 100)'
     )
     parser.add_argument(
         '--n-calib',

scripts/compute_fnr_table.py

@@ -97,8 +97,8 @@ def main():
     parser.add_argument(
         '--n-trials',
         type=int,
-        default=10,
-        help='Number of calibration trials (default: 10)'
+        default=100,
+        help='Number of calibration trials (default: 100)'
     )
     parser.add_argument(
         '--n-calib',

scripts/slurm_compute_fdr_thresholds.sh

@@ -5,7 +5,7 @@
 #SBATCH --ntasks=1
 #SBATCH --cpus-per-task=4
 #SBATCH --mem=32G
-#SBATCH --time=04:00:00
+#SBATCH --time=24:00:00
 #SBATCH --output=/groups/doudna/projects/ronb/conformal-protein-retrieval/logs/fdr_thresholds_%j.log
 #SBATCH --error=/groups/doudna/projects/ronb/conformal-protein-retrieval/logs/fdr_thresholds_%j.err
 
@@ -28,13 +28,27 @@ echo "Start time: $(date)"
 echo "Node: $(hostname)"
 echo ""
 
+# Exact match FDR
+echo "=== Computing EXACT match FDR thresholds ==="
 python scripts/compute_fdr_table.py \
     --calibration data/pfam_new_proteins.npy \
     --output results/fdr_thresholds.csv \
-    --n-trials 10 \
+    --n-trials 100 \
     --n-calib 1000 \
     --seed 42
 
+echo ""
+
+# Partial match FDR
+echo "=== Computing PARTIAL match FDR thresholds ==="
+python scripts/compute_fdr_table.py \
+    --calibration data/pfam_new_proteins.npy \
+    --output results/fdr_thresholds_partial.csv \
+    --n-trials 100 \
+    --n-calib 1000 \
+    --seed 42 \
+    --partial
+
 echo ""
 echo "============================================"
 echo "Completed: $(date)"

scripts/slurm_compute_fnr_thresholds.sh

@@ -27,21 +27,23 @@ echo "Start time: $(date)"
 echo "Node: $(hostname)"
 echo ""
 
-# Compute exact match FNR thresholds
+# Exact match FNR
 echo "=== Computing EXACT match FNR thresholds ==="
 python scripts/compute_fnr_table.py \
     --calibration data/pfam_new_proteins.npy \
     --output results/fnr_thresholds.csv \
-    --n-trials 10 \
+    --n-trials 100 \
     --n-calib 1000 \
     --seed 42
 
 echo ""
+
+# Partial match FNR
 echo "=== Computing PARTIAL match FNR thresholds ==="
 python scripts/compute_fnr_table.py \
     --calibration data/pfam_new_proteins.npy \
     --output results/fnr_thresholds_partial.csv \
-    --n-trials 10 \
+    --n-trials 100 \
     --n-calib 1000 \
     --seed 42 \
     --partial