Identification of Novel Inhibitors Targeting Inflammatory Pathways in Rheumatoid Arthritis through Drug Design, Computational Docking, and Prediction Studies

Table 1: Chemical structures and molecular formulas of the selected compounds.

Sr. No.	Compounds	Molecular formula	Structure
1	ortho-benzylphenols	C13H12O
2	3,4-dimethylphenol	C8H10O
3	para-cresol	C7H8O
4	2-(1-phenylethyl) phenol	C14H14O
5	2-methyl-6-(1-phenylethyl) phenol	C15H16O
6	4,5-dimethyl-2-(1-phenylethyl) phenol	C16H18O
7	2-chloro-6-(1-phenylethyl) phenol	C14H13ClO
8	4-ethyl-2-(1-phenylethyl) phenol	C16H18O
9	4-methyl-2-(1-phenylethyl) phenol	C84H90O4
10	para-fluorophenol	C6H5FO
11	ortho-benzylated phenols	C13H11ClO
12	3,5,7-trihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one	C15H10O6
13	3,5,7-Trihydroxy-2-(3,4,5-trihydroxyphenyl)-4H-1-benzopyran-4-one	C15H10O8
14	5,7-dihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one	C15H10O5
15	3',4',5,7-tetrahydroxyflavone	C15H10O6

2.2. Target Protein Selection

Two key therapeutic targets involved in the pathogenesis of rheumatoid arthritis (RA), tumor necrosis factor-alpha (TNF-α) and Janus kinase 3 (JAK3), were selected based on an extensive literature review due to their critical roles in regulating inflammatory signaling pathways. TNF-α is a major pro-inflammatory cytokine responsible for synovial inflammation and joint destruction, whereas JAK3 mediates cytokine-induced signaling through the JAK/STAT pathway, contributing to immune cell activation and disease progression. The three-dimensional crystal structures of TNF-α (PDB ID: 6RMJ) and JAK3 (PDB ID: 3LXK) were retrieved from the Protein Data Bank (PDB) and used for molecular docking studies.(19)

Target Protein 1 (TNF-α): The crystal structure of human tumor necrosis factor-alpha (TNF-α) was retrieved from the Protein Data Bank using PDB ID: 6RMJ.(20) This protein is classified as a cytokine, originates from Homo sapiens, and was expressed in Escherichia coli BL21 (DE3). TNF-α plays a central role in mediating inflammatory responses and is one of the primary therapeutic targets in rheumatoid arthritis.(21)

Target Protein 2 (JAK3): The crystal structure of Janus kinase 3 (JAK3) was obtained from the Protein Data Bank using PDB ID: 3LXK. JAK3 is classified as a transferase (protein kinase),(22) is of Homo sapiens origin, and was expressed in Spodoptera frugiperda. As a key component of the JAK/STAT signaling pathway, JAK3 regulates cytokine-mediated immune responses and is an important molecular target for developing anti-rheumatoid arthritis agents.(23)

2.3. Protein Preparation

The crystal structures of TNF-α (PDB ID: 6RMJ) and JAK3 (PDB ID: 3LXK) were downloaded from the Protein Data Bank (PDB) in PDB format. Before molecular docking, the protein structures were prepared by removing co-crystallized ligands, water molecules, and other non-essential heteroatoms. Missing hydrogen atoms were added, and the protein structures were optimized to obtain suitable receptor conformations for docking analysis.(24) The prepared proteins were subsequently used for molecular docking to evaluate the binding affinity and molecular interactions of the selected compounds.(25)

Figure 1: Crystal structure of human NGR-TNF, Crystal structure of protein JAK3 Kinase

2.4. Ligand preparation

A total of 15 phenolic and flavonoid compounds were selected based on their structural diversity and reported anti-inflammatory potential, identified through an extensive literature survey. The two-dimensional (2D) chemical structures were drawn using ChemDraw and subsequently converted into three-dimensional (3D) structures using Open Babel.(26)(27) The generated 3D conformations were subjected to geometry optimization and energy minimization to obtain stable ligand structures suitable for molecular docking studies. The optimized ligands were then used for subsequent docking analyses against the selected target proteins.(28)

2.5 Molecular Docking

Docking studies were conducted using the SwissDock web server, which uses the Dock DSS algorithm to predict binding conformations and estimate the free energy of binding.(29) Each ligand was docked against both 6RMJ and 3LXK to evaluate interaction profiles. The docking findings were studied based on Full Fitness and Estimated ΔG scores, with lower binding energies indicating stronger interactions.(29)(30)

1. 6. ADME Prediction

The pharmacokinetic properties of the lead compound identified through molecular docking were evaluated using the SwissADME web server. Parameters including absorption, distribution, metabolism, excretion, and drug-likeness were assessed to predict its pharmacokinetic behavior.(31) Physicochemical descriptors such as molecular weight, lipophilicity (LogP), topological polar surface area (TPSA), water solubility, gastrointestinal (GI) absorption, blood–brain barrier (BBB) permeability, P-glycoprotein (P-gp) substrate specificity, cytochrome P450 (CYP) enzyme inhibition, Lipinski's Rule of Five, bioavailability score, and medicinal chemistry parameters were recorded.(32) These predictions were used to evaluate the compound's suitability as a potential drug candidate.(33)

2.7. Toxicity Prediction

The toxicity profile of the lead compound was predicted using the ProTox-II web server. Various toxicity endpoints, including hepatotoxicity, carcinogenicity, mutagenicity, immunotoxicity, cytotoxicity, organ toxicity, Tox21 pathway analysis, molecular initiating events, and cytochrome P450-related toxicity, were evaluated.(34) The predicted toxicity data were used to assess the safety profile and potential toxicological risks of the selected compound before further experimental investigations.(35)

3. RESULT AND DISCUSSION

In this study, we explored how fifteen small molecules interact with TNF-α, a key driver of inflammation in rheumatoid arthritis, using molecular docking techniques. Our goal was to assess their potential as inhibitors that could block or weaken this inflammatory signal.

Among the tested compounds, 2-(1-phenylethyl) phenol showed the most promising interaction with TNF-α, reflected by its docking score of −5.86 kcal/mol, which suggests a strong and stable binding. Close behind were 3,4-dimethylphenol (−5.83 kcal/mol) and ortho-benzylphenols (-5.77 kcal/mol), indicating that these molecules may also have inhibitory potential. These results highlight the importance of small hydrophobic groups like methyl or phenylethyl, which seem to improve the molecule’s fit within the protein's active site.

On the other hand, some larger or highly polar compounds-especially flavonoid-based structures didn’t perform as well. Their docking scores were generally less favorable, ranging between −2.6 and −4.4 kcal/mol. This could be due to their bulkier shapes or the presence of multiple hydroxyl groups, which may hinder their ability to snugly fit into the protein’s binding pocket.

Interestingly, one compound -4-methyl-2-(1-phenylethyl) phenol returned an unusually high positive docking score (+24.36 kcal/mol), which typically indicates a failed interaction or docking error. This result was excluded from further interpretation.

Overall, the study suggests that smaller, moderately hydrophobic molecules tend to bind more effectively with TNF-α. These findings provide a useful direction for identifying potential anti-inflammatory drug candidates. In particular, 2-(1-phenylethyl) phenol stands out as a lead compound worthy of further investigation through ADME and toxicity studies to assess its potential as a drug for RA.

The structural analysis of the docked complexes also revealed meaningful insights into how these molecules interact with TNF-α at the molecular level. Most of the effective ligands formed favorable hydrophobic interactions and, in some cases, π-π stacking with key amino acid residues within the binding site. These interactions are essential for stabilizing the ligand-protein complex and could help prevent TNF-α from binding to its receptor. While hydrogen bonding was observed in some cases, it appeared less critical for binding affinity in this particular protein context. These findings not only support the docking scores but also reinforce the relevance of these interactions in guiding structure-based drug design. Taken together, the results highlight a clear direction for selecting and optimizing small-molecule inhibitors with desirable pharmacological profiles for the management of rheumatoid arthritis.

Table 2. showing SwissParam score for different ligands and protein TNF- alpha Human (NGR-TNF: 6RMJ)

1. 1. Comparative Docking Results with 3LXK

Following the promising interactions observed with TNF-α (6RMJ), the same fifteen compounds were docked against a second key RA-related protein target, 3LXK. This structure represents an additional node in the inflammatory signaling network, providing a broader perspective on the inhibitory potential of the compounds.

Interestingly, in contrast to the 6RMJ results, several flavonoid derivatives performed much better with 3LXK. Compounds like 3,5,7-Trihydroxy-2-(3,4,5-trihydroxyphenyl)-4H-1-benzopyran-4-one (−7.04 kcal/mol), 3',4',5,7-tetrahydroxyflavone (−6.93 kcal/mol), and 3,4',5,7-tetrahydroxyflavone (−6.90 kcal/mol) showed significantly improved binding energies compared to their previous interactions with TNF-α. This suggests a more favorable binding pocket or interaction mode in 3LXK for polyhydroxylated scaffolds.

Among phenolic derivatives, 4-methyl-2-(1-phenylethyl) phenol recorded an excellent docking score (−6.79 kcal/mol)—a stark contrast to its anomalous result with TNF-α. Similarly, 4-ethyl-2-(1-phenylethyl) phenol (−6.56 kcal/mol) and ortho-benzylated phenols (−6.50 kcal/mol) also demonstrated strong binding, pointing to a consistent potential across both protein targets.

Notably, the lead compound from the TNF-α study, 2-(1-phenylethyl) phenol, also maintained solid performance here with a docking score of −6.25 kcal/mol, reinforcing its candidacy for further optimization.

Table 3 showing SwissParam score for different ligands and protein 3LXK

3.1.1. ortho-benzylphenols

Figure 3 (A) showing protein (Human TLR2:  1FYW)  and ortho-benzylphenols (ligand) interaction and (B) binding pockets  along with hydrogen bond ionic interaction, Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Table 4 showing SwissParam score for ligand ortho-benzylphenols and protein TNF- alpha Human (NGR-TNF: 6RMJ)

3.1.2. 3,4-dimethylphenol

Figure 4: Chemical structure of 3,4-dimethylphenol and ligand parameter (box size and center) for target protein 6RMJ

Figure 5 (A) showing protein (Human TLR2:  1FYW)  and 3,4-dimethylphenol (ligand) interaction and (B) binding pockets  along with hydrogen bond ionic interaction, Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Table 5 showing SwissParam score for ligand 3,4-dimethylphenol and protein TNF- alpha Human (NGR-TNF: 6RMJ)

1. 1. 3. para-cresol

Figure 6 : Chemical structure of para-cresol and ligand parameter (box size and center) for target protein 6RMJ

Figure 7 (A) showing protein (Human TLR2:  1FYW)  and para-cresol (ligand) interaction and (B) binding pockets  along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Table 6 showing SwissParam score for ligand para-cresol and protein TNF- alpha Human (NGR-TNF: 6RMJ)

3.1.4 2-(1-phenylethyl) phenol

Figure 8: Chemical structure of 2-(1-phenylethyl) phenol and ligand parameter (box size and center) for target protein 6RMJ

Figure 9 (A) Showing protein (Human TLR2:  1FYW)  and 2-(1-phenylethyl) phenol (ligand) interaction and (B) binding pockets  along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, and Hydrophobic contacts and protein surface

Table 7 showing swissParam score for ligand 2-(1-phenylethyl) phenol and protein TNF- alpha Human (NGR-TNF: 6RMJ)

3.1.5 2-methyl-6-(1-phenylethyl) phenol

Figure 10: Chemical structure and ligand parameter (box size and center) for 2-methyl-6-(1-phenylethyl) phenol

Figure 11 (A) showing protein (Human TLR2:  1FYW)  and 2-methyl-6-(1-phenylethyl) phenol (ligand) interaction and (B) binding pockets  along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, hydrophobic contacts and protein surface

Table 8 showing swissParam score for ligand 2-methyl-6-(1-phenylethyl) phenol and protein TNF- alpha Human (NGR-TNF: 6RMJ)

3.1.6 4,5-dimethyl-2-(1-phenylethyl) phenol

Figure 12: Chemical structure and ligand parameter (box size and center) for 4,5-dimethyl-2-(1-phenylethyl) phenol

Figure 13 (A) showing protein (Human TLR2:  1FYW) and 4,5-dimethyl-2-(1-phenylethyl) phenol (ligand) interaction and (B) binding pockets along with hydrogen bond ionic interaction, Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Table 9 showing swissParam score for ligand 4,5-dimethyl-2-(1-phenylethyl) phenol and protein TNF- alpha Human (NGR-TNF: 6RMJ)

3.1.7 2-chloro-6-(1-phenylethyl) phenol

Figure 14: Chemical structure and ligand parameter (box size and center) for 2-chloro-6-(1-phenylethyl) phenol

Figure 19 (A) showing protein (Human TLR2:  1FYW) and 2-chloro-6-(1-phenylethyl) phenol (ligand) interaction, and (B) binding pockets along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface.

Table 10 showing swissParam score for ligand 2-chloro-6-(1-phenylethyl) phenol and protein TNF- alpha Human (NGR-TNF: 6RMJ)

3.1.8 4-ethyl-2-(1-phenylethyl) phenol

Figure 15: Chemical structure and ligand parameter (box size and center) for 4-ethyl-2-(1-phenylethyl) phenol

Figure 16 (A) showing protein (Human TLR2:  1FYW) and 4-ethyl-2-(1-phenylethyl) phenol (ligand) interaction, and (B) binding pockets along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Table 11 showing swissParam score for ligand 4-ethyl-2-(1-phenylethyl) phenol and protein TNF- alpha Human (NGR-TNF: 6RMJ)

3.1.9 4-methyl-2-(1-phenylethyl) phenol

Figure 17: Chemical structure and ligand parameter (box size and center) for 4-methyl-2-(1-phenylethyl) phenol

Figure 16 (A) showing protein (Human TLR2:  1FYW) and 4-ethyl-2-(1-phenylethyl) phenol (ligand) interaction, and (B) binding pockets along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Figure 18 (A) showing protein (Human TLR2: 1FYW) and 4-methyl-2-(1-phenylethyl) phenol (ligand) interaction, and (B) binding pockets along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Table 12 showing swissParam score for ligand 4-methyl-2-(1-phenylethyl) phenol and protein TNF- alpha Human (NGR-TNF: 6RMJ)

3.1.10 para-fluorophenol

Figure 19 : Chemical structure and ligand parameter (box size and center) for para-fluorophenol

Figure 20 (A) showing protein (Human TLR2:  1FYW) and para-fluorophenol (ligand) interaction and (B) binding pockets along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface.

3.1.11 ortho-benzylated phenols

Figure 21: Chemical structure and ligand parameter (box size and center) for ortho-benzylated phenols

Table 13 showing SwissParam score for ligand para-fluorophenol and protein TNF-alpha Human (NGR-TNF: 6RMJ)

Figure 22 (A) showing protein (Human TLR2:  1FYW) and ortho-benzylated phenols (ligand) interaction, and (B) binding pockets along with hydrogen bond ionic interaction, Cation-π interactions, π-stacking interactions, and Hydrophobic contacts and protein surface

Table 14 showing SwissParam score for ligand ortho-benzylated phenols and protein TNF- alpha Human (NGR-TNF: 6RMJ)

1. 1. 12. 3,5,7-trihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one or 3,4',5,7-tetrahydroxyflavone

Figure 23: Chemical structure and ligand parameter (box size and center) for 3,5,7-trihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one or 3,4',5,7-tetrahydroxyflavone

Figure 24 (A) showing protein (Human TLR2:  1FYW) and 3,5,7-trihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one or 3,4',5,7-tetrahydroxyflavone (ligand) interaction and (B) binding pockets along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, and Hydrophobic contacts and protein surface

Table 15 showing swissParam score for ligand 3,5,7-trihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one or 3,4',5,7-tetrahydroxyflavone and protein TNF- alpha Human (NGR-TNF: 6RMJ)

1. 1. 13. 3,5,7-Trihydroxy-2-(3,4,5-trihydroxyphenyl)-4H-1-benzopyran-4-one

Figure 25: Chemical structure and ligand parameter (box size and center) for 3,5,7-Trihydroxy-2-(3,4,5-trihydroxyphenyl)-4H-1-benzopyran-4-one

Figure 26 (A) showing protein (Human TLR2:  1FYW) and 3,5,7-Trihydroxy-2-(3,4,5-trihydroxyphenyl)-4H-1-benzopyran-4-one (ligand) interaction and (B) binding pockets  along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, and Hydrophobic contacts and protein surface

Table 16 showing swissParam score for ligand 3,5,7-Trihydroxy-2-(3,4,5-trihydroxyphenyl)-4H-1-benzopyran-4-one and protein TNF- alpha Human (NGR-TNF: 6RMJ)

1. 1. 14. 5,7-dihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one

Figure 27 : Chemical structure and ligand parameter (box size and center) for 5,7-dihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one

Figure 28 (A) showing protein (Human TLR2:  1FYW) and 5,7-dihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one (ligand) interaction and (B) binding pockets  along with hydrogen bond ionic interaction Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Table 17 showing swissParam score for ligand 5,7-dihydroxy-2-(4-hydroxyphenyl)-4H-chromen-4-one and protein TNF- alpha Human (NGR-TNF: 6RMJ)

1. 1. 15. 3',4',5,7-tetrahydroxyflavone

Figure 29: Chemical structure and ligand parameter (box size and center) for 3',4',5,7-tetrahydroxyflavone

Figure 30 (A) showing protein (Human TLR2:  1FYW) and 3',4',5,7-tetrahydroxyflavone (ligand) interaction and (B) binding pockets along with hydrogen bond ionic interaction, Cation-π interactions, π-stacking interactions, hydrophobic contacts, and protein surface

Table 18 showing SwissParam score for ligand 3',4',5,7-tetrahydroxyflavone and protein TNF- alpha Human (NGR-TNF: 6RMJ)

1. 2. ADME Profiling Prediction:

To evaluate the drug-likeness and pharmacokinetic feasibility of the top-performing compound, a detailed ADME analysis was carried out using SwissADME. The selected molecule (3,5,7-Trihydroxy-2-(3,4,5-trihydroxyphenyl)-4H-1-benzopyran-4-one), with a molecular weight of 318.24 g/mol, falls within an acceptable range for oral bioavailability. However, its topological polar surface area (TPSA) of 151.59 Ų is relatively high, which often correlates with limited passive absorption—a trend reflected in its predicted low gastrointestinal (GI) absorption.

In terms of lipophilicity, the compound exhibits a moderate consensus LogP value of 0.79, suggesting a balanced hydrophilic-lipophilic profile. This can be beneficial for systemic circulation, although the moderate to low predicted skin permeation (Log Kp −7.40 cm/s) indicates minimal transdermal potential. Notably, the compound is predicted not to cross the blood-brain barrier, which is advantageous for targeting peripheral inflammatory conditions like rheumatoid arthritis, as CNS side effects are minimized.

The compound shows favorable water solubility, with all three predictive models (ESOL, Ali, and SILICOS-IT) placing it in the “soluble” category. This supports its potential for formulation in aqueous-based delivery systems. From a metabolic standpoint, the molecule is not a substrate for P-glycoprotein (P-gp), minimizing the risk of efflux-related resistance. However, it may act as an inhibitor of CYP1A2 and CYP3A4, suggesting that caution is needed when co-administered with drugs metabolized by these enzymes.

From a drug-likeness perspective, the compound adheres to Lipinski’s Rule of Five with a single violation (due to its higher number of hydrogen bond donors). It also passes Ghose's filter but fails the Veber, Egan, and Muegge rules, largely due to its high polarity and hydrogen bonding potential. These features, while possibly reducing oral bioavailability, may enhance target specificity in inflammatory conditions.

The bioavailability score of 0.55 suggests moderate systemic exposure if administered orally. Medicinal chemistry filters flagged catechol substructures via both PAINS and Brenk alerts, which are known to sometimes produce false positives in bioassays. Despite this, the compound’s synthetic accessibility score of 3.27 indicates a reasonable path toward chemical synthesis and future optimization.

Overall, while some limitations exist—particularly in terms of permeability and enzyme inhibition—this compound presents a structurally promising lead for further refinement, especially as a selective, non-CNS-active JAK3 inhibitor in rheumatoid arthritis therapy.

Figure 31: The color zone is suitable physiochemical space for oral bioavailability

Table 19: ADME profile of selected ligand 2-(1-phenylethyl) phenol showing Physicochemical character, Water Solubility, Lipophilicity, Pharmacokinetics, Druglikeness,

1. 3. Toxicity Assessment of the Selected Molecule

To assess the safety and toxicological profile of the lead compound identified through docking—(3,5,7-Trihydroxy-2-(3,4,5-trihydroxyphenyl)-4H-1-benzopyran-4-one)—we conducted a detailed prediction using the ProTox-II server. This flavonoid-based molecule, structurally characterized by multiple hydroxyl groups and a polyphenolic backbone, showed a mixed toxicity profile with some areas of concern and others indicating acceptable safety.

From an organ toxicity standpoint, the molecule was predicted to be non-hepatotoxic (probability: 0.69) and non-neurotoxic (0.89), suggesting minimal risk to liver and nervous system tissues. However, nephrotoxicity (0.62) and respiratory toxicity (0.83) were flagged as potential issues. These findings imply a need for focused investigation on renal and pulmonary safety, particularly in chronic administration scenarios.

The model also indicated possible carcinogenic (0.68) and mutagenic (0.51) tendencies, which may be attributed to its polyhydroxylated structure. These alerts do not confirm genotoxicity but suggest that the molecule may influence pathways associated with genomic instability, warranting further experimental assessment.

Importantly, cytotoxicity and immunotoxicity risks were low, with respective probabilities of 0.99 and 0.86 for inactivity. This indicates a reduced likelihood of inducing general cell damage or compromising immune function, reinforcing the compound’s potential viability from a systemic toxicity standpoint.

In terms of molecular initiating events and receptor signaling:

• The compound was predicted to activate the aryl hydrocarbon receptor (AhR, 0.91) and both estrogen receptor alpha (ER, 0.87) and its ligand binding domain (ER-LBD, 0.95). These interactions suggest a potential for endocrine disruption, which could be problematic with prolonged exposure.
• On the other hand, no significant interaction was predicted for androgen receptors or PPARγ, which may limit broader hormonal side effects.

From a stress response perspective, activation of mitochondrial membrane potential (MMP) was observed (1.0), raising concerns about mitochondrial health and cellular respiration under drug exposure. However, the compound showed no significant activity on the p53 tumor suppressor or other key stress pathways like ATAD5, HSE, or IS, reducing concern for widespread oxidative or genotoxic stress.

Regarding metabolic enzyme interaction, the molecule showed potential to reduce crucial CYPP450 enzymes such as CYP1A2 (1.0), CYP2C19 (0.77), and CYP2C9 (0.99). This may imply a risk of drug-drug interactions, particularly when co-administered with other substrates metabolized via these pathways. However, it was predicted inactive against CYP3A4, CYP2D6, and CYP2E1, indicating selectivity in its metabolic impact.

Figure 32: Oral toxicity prediction results for input compound

Figure 33: Comparison of input compound with dataset compounds

Figure 34: The toxicity radar chart is to depict the confidence of positive toxicity findings.

Table 20: Toxicity Model Report

++ denoted active, --denoted as inactive

Figure 35. The network chart to illustrate the connection between the selected compound and predicted activitie

4. CONCLUSION