Nine immunology targets spanning kinases, checkpoint proteins, cytokines, and cell-surface antigens. Tests the engine's ability to discriminate between small-molecule and biologics modalities — the critical decision that determines drug discovery strategy. All results produced by the Ashebo engine from first-principles physics — no external database lookups.
All four kinases scored as PURSUE with scores ranging from 69 to 95. The ordering matches clinical maturity: JAK1 (95, multiple approved) > BTK (88, multiple approved) > TYK2 (82, one approved) > IRAK4 (69, clinical trials). The engine's score gradient tracks the clinical validation gradient.
PD-L1 scored 93/100 (97th percentile) with small-molecule potential confirmed by Layer E. This is the sprint's most commercially relevant finding. While PD-L1 is currently dominated by biologics, BMS has small-molecule PD-L1 inhibitors in clinical trials. The engine identifies the structural opportunity that the industry is beginning to exploit.
CTLA-4 scored 83/100 (structurally druggable) but Layer E correctly classified all leads as "Biologic Only." CD20 scored 74/100 but was similarly flagged. This demonstrates the engine's ability to separate structural druggability (Layer A) from therapeutic modality (Layer E) — a critical distinction that prevents false-positive pursue recommendations for biologics targets.
IL-17A scored 22/100, correctly identifying this homodimeric cytokine as undruggable for small molecules. TNF-alpha scored 66/100 (borderline) — reflecting the genuine ambiguity where small-molecule inhibitors exist (SPD-304) but none have been approved. The engine captures the nuance between "impossible" and "difficult."
| Metric | Oncology | Neurodegeneration | Immunology |
|---|---|---|---|
| Targets | 9 | 9 | 9 |
| Pursue | 2 (22%) | 5 (56%) | 5 (56%) |
| Borderline | 1 | 2 | 2 |
| No-Go | 4 (44%) | 2 (22%) | 2 (22%) |
| Mean Score | 52.3 | 60.8 | 74.7 |
| Highest | 94 (RB1) | 91 (LRRK2) | 95 (JAK1) |