Make every command runnable from the repo root; fix the folder-limit and gate wording

DATA.md

@@ -62,18 +62,20 @@ Every dataset ships the same four companion files, and all four are needed.
 
 ## graphs/
 
-Large datasets ship as tars because Hugging Face caps a folder at 10,000 entries.
+Large datasets ship as tars because Hugging Face documents a ceiling of 10,000 entries per folder.
 
 ### `graphs/tcga_brca/`: 11,149,499 graphs, 652 GB
 The **pre-training corpus**. 324 × ~2.0 GB tars under `tcga_brca/data/`, plus the four
 companion files (its `sample_labels_rich.csv` is 1.94 GB). Also the in-domain cohort for
 slide-level evaluation.
 
+Run from the repository root:
+
 ```bash
-mkdir -p tcga_brca/graphs
-for t in tcga_brca/data/tcga-brca-*.tar; do tar -xf "$t" -C tcga_brca/graphs; done
-python modeling/rebase_graph_paths.py --csv tcga_brca/sample_labels_rich.csv \
-       --mode absolute --root "$PWD/tcga_brca/graphs"
+mkdir -p graphs/tcga_brca/extracted
+for t in graphs/tcga_brca/data/tcga-brca-*.tar; do tar -xf "$t" -C graphs/tcga_brca/extracted; done
+python modeling/rebase_graph_paths.py --csv graphs/tcga_brca/sample_labels_rich.csv \
+       --mode absolute --root "$PWD/graphs/tcga_brca/extracted"
 ```
 
 ### `graphs/bach/`: 14,341 graphs
@@ -82,7 +84,9 @@ BACH (ICIAR 2018), 224 px @ 20×, 4-class. One `graphs.tar` (flat: the `.pt` fil
 create yourself:
 
 ```bash
-mkdir -p bach/graphs && tar -xf bach/graphs.tar -C bach/graphs
+mkdir -p graphs/bach/extracted && tar -xf graphs/bach/graphs.tar -C graphs/bach/extracted
+python modeling/rebase_graph_paths.py --csv graphs/bach/extracted/sample_labels_rich.csv \
+       --mode absolute --root "$PWD/graphs/bach/extracted"
 ```
 
 ### `graphs/bracs/`: 96,153 graphs
@@ -225,7 +229,7 @@ cells, columns `centroid_x_px,centroid_y_px,cell_type,type_prob`) plus a summary
 `code/` and `paper_metrics/` hold the analysis code and its outputs.
 
 The pseudo-labeller was validated on a held-out ground-truth split first: **ARI 0.560**
-(pre-registered threshold 0.5), Cohen κ 0.709, per-class confusion diagonal 0.864 / 0.562 /
+(threshold 0.5), Cohen κ 0.709, per-class confusion diagonal 0.864 / 0.562 /
 0.900 / 0.000. Three of four classes pass, and the fourth (`other_nucleus`) is untestable at
 7 ground-truth instances. Re-running that gate needs the external CellViT++ repository and
 SAM-H weights, which are not included here.
@@ -238,8 +242,7 @@ conclusion changes, but it needs a minimum-degree guard.
 ### `studies/adaptergnn/`: parameter-efficient fine-tuning
 
 Adapters inserted into a **frozen GrapHist v1** encoder for cell-type identification. Test
-macro-F1 %, ± is population std over CV folds (2–5 folds depending on the column, so the ± are
-not comparable across columns):
+macro-F1 %, each the mean over cross-validation folds (2 to 5 folds depending on the column):
 
 | Model | P20 Breast | P20 PanCancer | P40 Breast | P40 PanCancer | NuCLS main | NuCLS super |
 |---|:--:|:--:|:--:|:--:|:--:|:--:|