Donker Tree Ensemble Consensus
Generated by scripts/run_donker_tree_ensemble.py.
This consolidates the nine recovered Donker section-scoped consensus trees as partial textual-affinity hypotheses. It does not treat the result as demonstrated document-copy provenance.
Inputs
| Dataset | Witnesses | Internal nodes |
|---|---|---|
donker_ath_1cor |
21 | 19 |
donker_ath_2cor_titus |
21 | 19 |
donker_ath_acts |
24 | 21 |
donker_ath_acts_13_28 |
24 | 19 |
donker_ath_acts_1_12 |
22 | 15 |
donker_ath_catholicepistles |
14 | 12 |
donker_ath_heb |
18 | 15 |
donker_ath_paulineepistles |
21 | 19 |
donker_ath_rom |
19 | 11 |
Ensemble Methods
- Direct tree-distance consensus: convert each section tree to normalized pairwise leaf distances, then average pairwise distances across sections where both witnesses occur.
- Exemplar-kernel consensus: for each section tree, repeatedly sample 4 exemplar witnesses for 500 trials and compare witnesses by their distance profiles to those exemplars.
The direct and exemplar-kernel pairwise distances have Spearman rho 0.9413 (p=1.85e-230) on observed pairs.
Outputs
- Direct consensus tree:
outputs/figures/donker_tree_ensemble/direct_tree_distance/consensus_provenance_tree.png - Exemplar-kernel consensus tree:
outputs/figures/donker_tree_ensemble/exemplar_kernel/consensus_provenance_tree.png - Direct distance matrix:
outputs/tables/donker_tree_ensemble/direct_tree_distance_matrix.csv - Exemplar-kernel distance matrix:
outputs/tables/donker_tree_ensemble/exemplar_kernel_distance_matrix.csv - Pairwise direct summary:
outputs/tables/donker_tree_ensemble/direct_pairwise_summary.csv - Pairwise exemplar summary:
outputs/tables/donker_tree_ensemble/exemplar_kernel_pairwise_summary.csv
In the tree PNGs, internal-node labels such as 4/9 report mean pairwise section coverage for that aggregate clade, not historical manuscript support.
Closest Stable Pairs
| left | right | section_count | mean_normalized_tree_distance | sd_normalized_tree_distance |
|---|---|---|---|---|
| F | G | 3 | 0.1513 | 0.0168 |
| M1704 | M945 | 3 | 0.1560 | 0.0392 |
| A | P74 | 3 | 0.1768 | 0.0300 |
| M1352 | P | 3 | 0.1768 | 0.0300 |
| C | M1175 | 3 | 0.2185 | 0.0863 |
| H | U49 | 3 | 0.2185 | 0.0863 |
| D | F | 3 | 0.2269 | 0.0252 |
| M223 | M876 | 5 | 0.2314 | 0.0699 |
| D | G | 4 | 0.2327 | 0.0236 |
| L | U49 | 7 | 0.2331 | 0.1117 |
| M223 | M2423 | 5 | 0.2336 | 0.1519 |
| M1704 | M630 | 3 | 0.2339 | 0.0588 |
Most Section-Variable Pairs
| left | right | section_count | mean_normalized_tree_distance | sd_normalized_tree_distance |
|---|---|---|---|---|
| P | U1C | 5 | 0.4613 | 0.2791 |
| Ath | P | 8 | 0.5740 | 0.2751 |
| D | M614 | 3 | 0.3911 | 0.2720 |
| C | U44 | 8 | 0.5984 | 0.2682 |
| U1 | U44 | 9 | 0.5735 | 0.2517 |
| B | M614 | 3 | 0.5375 | 0.2459 |
| M614 | U1 | 3 | 0.5375 | 0.2459 |
| D | H | 3 | 0.5613 | 0.2439 |
| D | U44 | 8 | 0.4708 | 0.2438 |
| U1C | U44 | 5 | 0.5102 | 0.2422 |
| B | H | 3 | 0.7077 | 0.2409 |
| H | U1 | 3 | 0.7077 | 0.2409 |
Consensus Newick
Direct tree-distance consensus:
((((((P46,M33),U1C),(D,(F,G))),(((M223,M876),M2423),(K,M104))),(((((A,P74),(Ath,M81)),((B,U1),(C,M1175))),(((M1704,M945),M630),M1891)),(((E,U44),(M383,M614)),(((H,U49),(L,M1073)),(M1352,P))))),((M1739,M323),((M1022,M1424),(M105,M201))));
Exemplar-kernel consensus:
((((((P46,M33),U1C),(F,G)),(((M223,M2423),M876),((K,M104),U44))),((((((A,P74),M81),(B,U1)),(C,M1175)),((Ath,M1891),((M1704,M945),M630))),(((D,E),(M383,M614)),(((H,U49),(L,M1073)),(M1352,P))))),((M1739,M323),((M1022,M1424),(M105,M201))));
Node Tables
- Direct aggregate nodes:
outputs/tables/donker_tree_ensemble/direct_consensus_nodes.csv(37 nodes) - Exemplar aggregate nodes:
outputs/tables/donker_tree_ensemble/exemplar_kernel_consensus_nodes.csv(37 nodes)