From Large Chemical Space to Candidates for Experimental Review
Graphical abstract
Summary
A staged funnel combines model ranking, structural review, scaffold diversity, developability risk and dynamic evidence to narrow a broad chemical space. The graphical abstract brings together candidate distributions, structural review and property comparisons across the screening funnel. The results establish a clear candidate hierarchy for experimental selection and subsequent lead optimisation. The result-focused presentation supports efficient review of the main evidence and research priorities.
Computational results
Statistical analysis of the results of cross-screening
Figure 2 summarizes the overall distributions and relationships of candidate compounds obtained during virtual-screening docking. In (A), Docking score is concentrated around −6 to −3 kcal/mol, with a mean of −4.05 and median of −4.07, indicating a near-symmetric distribution without substantial extreme-value bias. DrugCLIP score (B) is mainly within 0.9–1.2; its mean (0.98) is slightly higher than its median (0.95), suggesting that the AI model distinguishes candidate scores and identifies a subset of high-scoring molecules. In (C–D), molecular weight is mainly 300–500 Da (mean approximately 411.8 Da), and AlogP is mainly 2–4 (mean approximately 2.85), indicating moderate hydrophobicity and favorable drug-likeness.
This figure presents the principal structures and trends in “Statistical analysis of the results of cross-screening” and connects them to the case-level ranking and result interpretation.
The results of the virtual screening of candidate molecules
Scroll horizontally when the table is wider than the page
| Sequence A_ID | docking score | drugclip score | QED |
|---|---|---|---|
| Compound K | -7.892 | 0.97 | 0.848793 |
| Compound L | -7.445 | 0.97 | 0.775883 |
| Compound M | -7.251 | 0.88 | 0.724759 |
| Compound N | -7.048 | 1.1 | 0.738398 |
| Compound O | -6.863 | 1.31 | 0.709938 |
| Compound P | -6.812 | 1.21 | 0.713136 |
| Compound Q | -6.8 | 0.94 | 0.865378 |
| Compound R | -6.665 | 1.15 | 0.83075 |
| Compound S | -6.646 | 0.88 | 0.711086 |
| Compound T | -6.616 | 1.07 | 0.804729 |
| Compound U | -6.564 | 0.93 | 0.778714 |
| Compound V | -6.344 | 0.87 | 0.711086 |
| Compound W | -5.784 | 0.97 | 0.760226 |
| Compound X | -5.654 | 0.95 | 0.706078 |
| Compound Y | -5.528 | 0.87 | 0.707189 |
| Compound Z | -5.438 | 1.07 | 0.830711 |
| Compound A | -5.35 | 0.99 | 0.796518 |
| Compound B | -5.303 | 1.02 | 0.746018 |
| Compound C | -5.275 | 0.9 | 0.726631 |
| Compound D | -5.173 | 1 | 0.788227 |
| Compound E | -5.155 | 1.07 | 0.81681 |
| Compound F | -5.149 | 0.88 | 0.869749 |
| Compound G | -5.136 | 1.12 | 0.840967 |
| Compound H | -4.926 | 0.93 | 0.737217 |
| Compound I | -4.864 | 0.92 | 0.825447 |
| Compound J | -4.784 | 0.96 | 0.792325 |
| Compound K | -4.583 | 0.9 | 0.707443 |
| Compound L | -4.534 | 1.03 | 0.826844 |
| Compound M | -4.517 | 0.98 | 0.773463 |
| Compound N | -4.496 | 0.97 | 0.797064 |
| Compound O | -4.387 | 1.12 | 0.777253 |
| Compound P | -4.31 | 0.98 | 0.725644 |
| Compound Q | -4.291 | 0.86 | 0.830025 |
| Compound R | -4.267 | 1.02 | 0.754305 |
| Compound S | -4.256 | 0.97 | 0.787584 |
| Compound T | -4.175 | 0.88 | 0.799739 |
| Compound U | -4.144 | 0.94 | 0.729009 |
| Compound V | -4.05 | 0.89 | 0.885888 |
| Compound W | -3.679 | 0.91 | 0.791234 |
This table consolidates the key comparisons in “The results of the virtual screening of candidate molecules”, making differences across conditions and candidates directly reviewable.
Statistical analysis of virtual screening candidate molecules
Figure 3. Comprehensive statistical analysis of the 39 final candidate molecules. (A) Docking-score histogram; red and blue dashed lines indicate the mean (−5.440 kcal/mol) and median (−5.173 kcal/mol), respectively. (B) Docking score versus DrugCLIP score, with point color indicating QED. (C) Top 10 candidate molecules ranked by Docking score. (D) Box-plot analysis of QED across Docking-score ranges.
This figure presents the principal structures and trends in “Statistical analysis of virtual screening candidate molecules” and connects them to the case-level ranking and result interpretation.
Figure 5. Based on the pairing obtained: A) Target A_Compound M;; B) TargetA_Compaund L;; C) Targeta_Compuund N;; D) Target's binding mode: A_compound K, left to the overall view, right to the local view, yellow stick to the small molecule, blue cartoon to the protein, green to the hydrogen bonds, gray to the drainage, and yellow to the salt bridges.
This figure presents the principal structures and trends in “Statistical analysis of virtual screening candidate molecules” and connects them to the case-level ranking and result interpretation.
Full case PDF
Complete results, high-resolution figures and analysis are provided in the client PDF. Add our service account on WeChat and send HS-CASE-0001 to request it.