[18:43:56] loading graph...
[18:43:57] N=47031 nodes, 2250197 directed edge rows, 24 metaedge types
[18:43:57] gene candidate pool size = 20945
[18:43:58] A_full nnz=4215418
[18:43:58] OA positive DaG genes = 62 (expect 62)
[18:43:59] weighted matrices built in 3.3s
[18:44:00] dwpc score vectors ready
[18:44:00] Part A: leave-one-out evaluation of arms (auroc/auprc/recall/mrr)
[18:44:47] seed 0 done: scorer-ppr:auroc=0.959, scorer-dwpc:auroc=0.937, scorer-adamic-adar:auroc=0.924, baseline-degree-null:auroc=0.758
[18:45:36] seed 1 done: scorer-ppr:auroc=0.959, scorer-dwpc:auroc=0.937, scorer-adamic-adar:auroc=0.924, baseline-degree-null:auroc=0.758
[18:46:26] seed 2 done: scorer-ppr:auroc=0.959, scorer-dwpc:auroc=0.937, scorer-adamic-adar:auroc=0.924, baseline-degree-null:auroc=0.758
[18:46:26] Part B: 36-setting grid (full graph, no fold removal) for Jaccard/Kendall robustness
[18:46:28] grid computed: 36 settings
[18:46:28] pairwise settings compared = 630 (expect C(36,2)=630); jaccard median=0.111, kendall median=0.503
[18:46:28] degree-null canonical AUROC=0.758; top50 overlap with pipelines={'scorer-ppr': 0, 'scorer-dwpc': 5, 'scorer-adamic-adar': 0} (max frac=0.10)
[18:46:28] TOTAL runtime = 152.4s
[18:46:28] results written to results.json
[18:46:28] per-fold raw ranks written to logs/per_fold_raw.jsonl (186 records)
[18:46:28] grid top50 lists written to logs/grid_top50.json
<workdir>/work/make_figs.py:56: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_xticklabels(names2, rotation=20, ha="right", fontsize=8)
wrote figs/summary.png