The DTSM project compound library has been profiled across three kinase assays: EGFR (Epidermal Growth Factor Receptor), PKCα (Protein Kinase C alpha), and CDK2 (Cyclin-Dependent Kinase 2). All assay values are reported as percentage activity where higher values indicate greater potency. The dataset covers 9,992 unique DTSM compounds, each fully profiled in all three assays. CDK2 shows the broadest dynamic range (min: -3, max: 1130), confirming strong and diverse hit matter within this series.
| Assay | Mean | Median | Std Dev | Min | Max | Inactive (<80) | Low (80–100) | Moderate (100–120) | High (≥120) |
|---|---|---|---|---|---|---|---|---|---|
| EGFR | 96.4 | 98.5 | 17.5 | -0.3 | 196.6 | 1,253 (12.5%) | 4,170 (41.7%) | 4,112 (41.2%) | 457 (4.6%) |
| PKCα | 101.5 | 102.5 | 37.6 | -215.0 | 1346.6 | 1,461 (14.6%) | 2,932 (29.3%) | 4,106 (41.1%) | 1,493 (14.9%) |
| CDK2 | 95.4 | 87.0 | 47.7 | -3.4 | 1129.9 | 3,529 (35.3%) | 3,822 (38.3%) | 1,153 (11.5%) | 1,488 (14.9%) |
There is a modest positive trend between molecular weight and CDK2 activity. Compounds in the 350–550 Da range show the highest median CDK2 activity (~90–104 units) compared to lighter molecules (<250 Da, median 82.6). This suggests that additional molecular complexity generally benefits CDK2 engagement, though the top CDK2 hits span a wide MW range (200–593 Da), indicating MW alone is not the primary driver.
A mild positive correlation exists between cLogP and CDK2 activity. Compounds with cLogP 4–6 show median CDK2 of 91.0, and >6 show 92.3, versus 84.4 for the 0–2 range. This is consistent with increased hydrophobic contacts in the CDK2 ATP-binding pocket, though many highly potent CDK2 hits carry cLogP < 2 — indicating hydrogen-bond-rich scaffolds can also achieve excellent CDK2 activity.
No significant correlation was observed between PSA and CDK2 activity, suggesting the CDK2 binding mode accommodates a wide polarity range. Several drug-like top hits fall in the 50–110 Ų window, which is expected to support good cell permeability.
Very low correlation between CDK2 and EGFR activity confirms that EGFR inhibition is largely independent of CDK2 activity across this library. CDK2-active hits are not systematically EGFR-active, making selectivity readily achievable. Only 19 compounds (0.19%) were pan-active across all three assays.
| MW Range | N | Median CDK2 | Mean CDK2 | Trend |
|---|---|---|---|---|
| <250 Da | 2,358 | 82.6 | 86.4 | |
| 250–350 Da | 3,814 | 86.8 | 93.8 | |
| 350–450 Da | 2,347 | 90.3 | 100.4 | |
| 450–550 Da | 1,008 | 89.6 | 104.4 | |
| >550 Da | 465 | 92.3 | 110.1 |
| cLogP Range | N | Median CDK2 | Trend |
|---|---|---|---|
| <0 | 759 | 87.1 | |
| 0–2 | 2,137 | 84.4 | |
| 2–4 | 3,976 | 85.4 | |
| 4–6 | 2,304 | 91.0 | |
| >6 | 816 | 92.3 |
Compounds were classified using a threshold of ≥120 activity units per assay, representing the top ~15% per assay.
1,064 compounds (10.6%) are CDK2-selective with no significant EGFR or PKCα cross-reactivity — a strong and drug-friendly hit rate. The near-equal proportion of CDK2-selective and PKCα-selective compounds, combined with only 19 pan-active (0.19%), confirms that the DTSM series has excellent capacity for kinase selectivity optimisation.
Selected via a two-stage algorithm: (1) ranking by combined CDK2 activity + selectivity score, then (2) greedy maximum-diversity filter using Morgan fingerprint Tanimoto similarity (threshold <0.40), ensuring structural diversity. Only compounds with CDK2 ≥ 120 and MW ≤ 700 Da were considered (1,476 candidates).
≥300 Very High CDK2 120–299 High CDK2 ≥120 Off-target concern (≥120) <120 Acceptable off-target