LigGen-a GEN-AI based ligand generation approach for de-novo drug design.

Code ↔ Paper

12 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 12 matches · 1 of them tie a paragraph to a whole file, not to given lines: a weak match, whose lines are not tinted

[1] § Results and discussion › Crossdocked dataset ↔ server/frontend/src/App.js, lines 665–729 · score 0.97 · DecompDiff, DiffSBDD, GraphBP, Lingo3DMol, TargetDiff, Frag2Seq
[2] § Results and discussion › Crossdocked dataset ↔ server/frontend/src/otherTools.js, the whole file · a weak match · score 0.92 · DecompDiff, DiffSBDD, GraphBP, Lingo3DMol, TargetDiff, Frag2Seq
[3] § Results and discussion › Comparative analysis of LigGen for alzheimer’s-related targets: BACE1, MAO-B, and AChE ↔ server/frontend/src/App.js, lines 665–729 · score 0.91 · AutoDock, LigBuilder, Pocket2Mol, Acetylcholinesterase, Beta, Monoamine
[4] § Methods › Recurrent neural network ↔ train_fragment_generator.py, lines 88–153 · score 0.73 · fragment generation, generative model, PyTorch, hidden, layer, mapping
[5] § Methods › Recurrent neural network ↔ rnn_selfies.py, lines 7–64 · score 0.69 · PyTorch, linear, Embedded, GRU, masking, SELFIES
[6] § Results and discussion ↔ server/frontend/src/App.js, lines 561–623 · score 0.66 · CrossDock, LigBuilder, LigGen, alzheimer, V3, novo
[7] § Methods › Fragments to ligands ↔ mcsa.py, lines 270–389 · score 0.63 · random fragment, vina score, rejected, position, conformation, optimizers
[8] § Methods › Fragments to ligands ↔ server/frontend/src/App.js, lines 374–434 · score 0.61 · vina score weight, synthesizability score, Metropolis, temperature, alpha, probability
[9] § Methods › Recurrent neural network ↔ train_fragment_generator.py, lines 88–153 · score 0.59 · fragment generation, generative model, trained, SELFIES, token, RNN
[10] § Methods › Fragments to ligands ↔ mcsa.py, lines 394–448 · score 0.59 · vina score weight, 0–1, rejection, alpha, RDKit, probability
[11] § Methods › Recurrent neural network ↔ rnn_selfies.py, lines 7–64 · score 0.54 · end token, start token, SELFIES, RNN, model
[12] § Methods › Fragments to ligands ↔ generate_ligands.py, lines 86–192 · score 0.51 · vina score, generated ligand, pose, box, RNN, SMILES

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The paper is loaded by your browser, with JavaScript: read it at doi.org or on Europe PMC.

The authors' code

JavaScript · 766 lines · 26 KB · no license · 4 matches

import React, { useState } from "react";
import OtherTools from "./otherTools";
import exampleFileContent from "./ExampleFile";
import {
AppBar,
TextField,
Checkbox,
MenuItem,
Button,
Typography,
Select,
FormControl,
FormControlLabel,
InputLabel,
Stack,
Box,
Container,
Paper,
Table,
TableHead,
TableRow,
TableCell,
TableBody,
Dialog,
DialogActions,
DialogContent,
DialogTitle,
Tabs,
Tab,
Toolbar,
TableContainer
} from "@mui/material";
import axios from "axios";
var ADDRESS = "https://neurocare-liggen.iiitd.edu.in/"
//var ADDRESS = "http://localhost:3000"
const App = () => {
const [jobId, setJobId] = useState("");
const [status, setStatus] = useState("");
const [file, setFile] = useState(null);
const [helpOpen, setHelpOpen] = useState(false);
const [activeTab, setActiveTab] = useState(0);
const handleTabChange = (event, newValue) => {
setActiveTab(newValue);
};
const handleHelpOpen = () => {
setHelpOpen(true);
};
const handleHelpClose = () => {
setHelpOpen(false);
};
const [parameters, setParameters] = useState({
count: 10,
grid_center: "",
grid_size: "",
threads: 1,
rnn_device: "gpu",
alpha: 0.3,
chain_extend_probability: 0.8,
weight: 500,
temp: 50,
score: 0,
vina_weight: 0.5,
jobId: "",
example: false
});
const [ligandDetailsDict, setLigandDetailsDict] = useState(null);
// Fetch ligand details from server
const fetchLigandDetails = async (jobId) => {
try {
const response = await axios.get(`${ADDRESS}/ligand/details/${parameters.jobId}`);
setLigandDetailsDict(response.data);
console.log(response.data['0']['img'])
} catch (error) {
console.error("Error fetching ligand details:", error);
}
};
const handleChange = (e) => {
const { name, value, type, checked } = e.target;
setParameters({
...parameters,
[name]: type === "checkbox" ? checked : value,
});
};
const submitJob = async () => {
let rnn_device = parameters.rnn_device;
if (rnn_device === "cpu") {
rnn_device = "cpu";
} else {
rnn_device = "cuda";
}
try {
let content = {
count: +parameters.count,
grid_center: parameters.grid_center.replaceAll(" ", "").split(',').map(str => parseFloat(str)),
grid_size: parameters.grid_size.replaceAll(" ", "").split(',').map(str => parseFloat(str)),
threads: +parameters.threads,
rnn_device: rnn_device,
alpha: parseFloat(parameters.alpha),
chain_extend_probability: parseFloat(parameters.chain_extend_probability),
weight: parseFloat(parameters.weight),
temp: parseFloat(parameters.temp),
score: parseFloat(parameters.score),
vina_weight: parseFloat(parameters.vina_weight),
target: file.content
}
const response = await axios.post(ADDRESS + "/submit-job", content);
setJobId(response.data.job_id);
} catch (error) {
console.error("Error submitting job:", error);
}
};
const checkStatus = async () => {
if (!parameters.jobId) return;
try {
const response = await axios.get(`${ADDRESS}/job-status/${parameters.jobId}`);
setStatus(response.data.status);
console.log("sdssfdsfdsfsdfsdfsdfds")
fetchLigandDetails();
} catch (error) {
console.error("Error checking job status:", error);
}
};
const handleDownload = () => {
if (!parameters.jobId) return;
// Construct the URL with the job ID
const downloadUrl = `${ADDRESS}/download/${parameters.jobId}`;
// Create an anchor element to programmatically trigger a download
const link = document.createElement("a");
link.href = downloadUrl;
link.download = `ligands_${parameters.jobId}.zip`; // Suggested file name
link.click();
};
// Handle file upload
const handleFileChange = (event) => {
const uploadedFile = event.target.files[0];
const reader = new FileReader();
reader.onload = (e) => {
const content = e.target.result;
setFile({
file_name: uploadedFile.name,
content: content
});
};
reader.readAsText(uploadedFile);
};
const loadExampleJob = async () => {
// 1. Pre-fill the parameters so the user can see what inputs generate these results
setParameters({
count: 10,
grid_center: "-4, -4, 30", // Replace with your actual 7D9O coordinates
grid_size: "30, 30, 30", // Replace with your actual 7D9O grid size
threads: 1,
rnn_device: "gpu",
alpha: 0.3,
chain_extend_probability: 0.8,
weight: 500,
temp: 50,
score: 0,
vina_weight: 0.5,
example: true
});
setFile({
file_name: "example_2g94.pdbqt",
content: exampleFileContent
});
};
const submitJobUI = () => {
return (<Container>
<Box sx={{ textAlign: "center", mt: 4 }}>
<Typography variant="subtitle1" color="textSecondary">
Submit jobs for ligand generation.
</Typography>
</Box>
<Box sx={{ mt: 4 }}>
<Stack direction="row" justifyContent="space-between">
<Typography variant="h5" gutterBottom>
Job Parameters
</Typography>
<Stack direction="row" spacing={2}>
<Button
variant="outlined"
color="secondary"
onClick={handleHelpOpen}
>
Help
</Button>
<Button
variant="outlined"
color="secondary"
onClick={loadExampleJob}
>
Example
</Button>
</Stack>
</Stack>
<br />
<Stack spacing={3}>
<Button
variant="contained"
component="label"
sx={{ textAlign: "center" }}
>
Upload Target Protein PDBQT
<input
type="file"
hidden
onChange={handleFileChange}
accept=".pdbqt"
/>
</Button>
{file && <Typography>Target File: {file.file_name}</Typography>}
<Stack direction="row" spacing={2}>
<TextField
fullWidth
label="Grid Center x, y, z"
name="grid_center"
value={parameters.grid_center}
onChange={handleChange}
/>
<TextField
fullWidth
label="Grid Size x, y, z"
name="grid_size"
value={parameters.grid_size}
onChange={handleChange}
/>
</Stack>
<Stack direction="row" spacing={2}>
<TextField
fullWidth
type="number"
label="Weight of Ligands"
name="weight"
value={parameters.weight}
onChange={handleChange}
/>
</Stack>
<Stack direction="row" spacing={2}>
<TextField
fullWidth
type="number"
label="Number of Ligands"
name="count"
value={parameters.count}
onChange={handleChange}
/>
<TextField
fullWidth
type="number"
label="Threads"
name="threads"
value={parameters.threads}
onChange={handleChange}
/>
</Stack>
<Stack direction="row" spacing={2}>
<TextField
fullWidth
type="number"
label="Temperature"
name="temp"
value={parameters.temp}
onChange={handleChange}
/>
<FormControl fullWidth>
<InputLabel>Device</InputLabel>
<Select
name="rnn_device"
value={parameters.rnn_device}
onChange={handleChange}
>
<MenuItem value="cpu">CPU</MenuItem>
<MenuItem value="gpu">GPU</MenuItem>
</Select>
</FormControl>
</Stack>
<Stack direction="row" spacing={2}>
<TextField
fullWidth
type="number"
step="0.1"
label="Vina Weight"
name="vina_weight"
value={parameters.vina_weight}
onChange={handleChange}
/>
<TextField
fullWidth
type="number"
step="0.1"
label="Alpha"
name="alpha"
value={parameters.alpha}
onChange={handleChange}
/>
</Stack>
<Stack direction="row" spacing={2}>
<TextField
fullWidth
type="number"
step="0.1"
label="Chain Extend Probability"
name="chain_extend_probability"
value={parameters.chain_extend_probability}
onChange={handleChange}
/>
<TextField
fullWidth
type="number"
step="1"
label="Score"
name="score"
value={parameters.score}
onChange={handleChange}
/>
</Stack>
</Stack>
<br />
<Stack>
<Button
variant="contained"
color="primary"
onClick={submitJob}
sx={{ mr: 2 }}
>
Submit Job
</Button>
</Stack>
</Box>
<Dialog open={helpOpen} onClose={handleHelpClose} fullWidth
maxWidth="md">
<DialogTitle>Parameter Descriptions</DialogTitle>
<DialogContent>
<Typography variant="body1" gutterBottom>
<strong>Grid Center x, y, z:</strong> Coordinates of the grid center for ligand generation.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Grid Size x, y, z:</strong> Dimensions of the grid for ligand generation.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Weight of Ligands:</strong> Weight in atomic mass units.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Number of Ligands:</strong> Number of ligands to generate.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Threads:</strong> Number of threads for computation.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Temperature:</strong> Starting temperature for the Metropolis criteria.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Device:</strong> Select CPU or GPU for computation.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Vina Weight:</strong> Weight of vina score between [0 to 1], Final score of ligand = (vina_score_weight)*vina_score + (1-vina_score_weight)*synthesizability_score
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Alpha:</strong> Factor by which cooling schedule (Temperature) changes.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Chain Extend Probability:</strong> Probablity by which fragment will get added to ends of ligand.
</Typography>
<Typography variant="body1" gutterBottom>
<strong>Score:</strong> Initial minimum score to accept a ligand.
</Typography>
</DialogContent>
<DialogActions>
<Button onClick={handleHelpClose} color="primary">
Close
</Button>
</DialogActions>
</Dialog>
</Container>);
}
const checkStatusUI = () => {
return (<Container sx={{ mt: 4 }}>
<Stack direction="row" spacing={2}>
<Button
variant="outlined"
color="secondary"
onClick={checkStatus}
>
Check Status
</Button>
<Button
variant="outlined"
color="secondary"
onClick={handleDownload}
>
Download Ligands
</Button>
<TextField
fullWidth
label="Job Id"
name="jobId"
value={parameters.jobId}
onChange={handleChange}
/>
</Stack>
</Container>);
}
// Render ligand details
const ligandDetailsUI = (ligandDetails, key) => {
console.log(ligandDetails);
if (!ligandDetails || Object.keys(ligandDetails).length === 0) return (
<Box sx={{ mt: 4 }}>
<Typography variant="h5" gutterBottom>
Ligand {key} : Generating
</Typography>
</Box>
);
return (
<Box sx={{ mt: 4 }}>
<Typography variant="h5" gutterBottom>
Ligand {key}
</Typography>
{/* General Details */}
<Paper sx={{ p: 2, mb: 4 }}>
{ligandDetails.img ? <Stack direction="row" spacing={2}>
{<img
src={`data:image/png;base64,${ligandDetails.img}`}
alt="Ligand"
style={{ border: "1px solid #ccc" }}
/>}
<Box>
<Typography>
<strong>Vina Score:</strong>{" "}
{ligandDetails.undocked_final_energy}
</Typography>
<Typography>
<strong>Synthesizability Score:</strong>{" "}
{ligandDetails.synthesizability_score}
</Typography>
</Box>
</Stack> : <Box> Generating </Box>}
</Paper>
{/* State Details Table */}
<Table>
<TableHead>
<TableRow>
<TableCell>Step</TableCell>
<TableCell>Added Fragment</TableCell>
<TableCell>Sub Ligand</TableCell>
<TableCell>Total Score</TableCell>
<TableCell>Vina Score</TableCell>
<TableCell>Synthesizability Score</TableCell>
</TableRow>
</TableHead>
<TableBody>
{ligandDetails.state_details.map((detail, index) => (
<TableRow key={index}>
<TableCell>{index + 1}</TableCell>
<TableCell>{detail.added_frag}</TableCell>
<TableCell>{detail.out_ligand}</TableCell>
<TableCell>{detail.total_score}</TableCell>
<TableCell>{detail.vina_score}</TableCell>
<TableCell>{detail.sa_score}</TableCell>
</TableRow>
))}
</TableBody>
</Table>
</Box>
);
};
const introductionUI = () => (
<Container maxWidth="md">
{/* Header */}
<Box sx={{ marginBottom: 4, textAlign: "center" }}>
<Typography variant="h4" component="h1" gutterBottom>
LigGen - A GEN-AI and Monte Carlo Simulated Annealing Based
De Novo Drug Design Toolbox
</Typography>
</Box>
{/* Introduction */}
<Box sx={{ marginBottom: 3 }}>
<Typography variant="body1" paragraph>
We have developed a novel Gen-AI and Monte Carlo Simulated Annealing
based de novo ligand generation tool referred to as <strong>LigGen</strong>, which
uses fragments in SMILES format as building blocks.
</Typography>
<Typography variant="body1" paragraph>
When compared to other de novo drug discovery packages such as
<strong> LigBuilder</strong> and <strong>Pocket2Mol</strong>, LigGen's fragment generation
is based on generative AI exploiting an RNN-LSTM approach. This approach
was pretrained on the <strong>ChEMBL dataset</strong> and fine-tuned on LigBuilder’s
fragment library.
</Typography>
</Box>
{/* Performance Highlights */}
<Box sx={{ marginBottom: 3 }}>
<Typography variant="h5" component="h2" gutterBottom>
Performance Highlights
</Typography>
<Typography variant="body1" paragraph>
LigGen’s performance was evaluated on the <strong>CrossDock dataset</strong> and
three Alzheimer’s-related proteins to compare it with several
state-of-the-art de novo drug design tools, including
<strong> Pocket2Mol</strong>, <strong>LigBuilder V3</strong>, and others.
</Typography>
<Typography variant="body1" paragraph>
On the CrossDock dataset, LigGen demonstrated competitive binding
affinities, achieving a Vina score of <strong>-7.508 ± 1.83</strong>,
outperforming baseline methods like <strong>3D-SBDD</strong>, <strong>FLAG</strong>, and
<strong>DrugGPS</strong>. LigGen also showcased strong diversity in the generated
molecules while maintaining reasonable synthesizability.
</Typography>
<Typography variant="body1" paragraph>
Although tools like Pocket2Mol showed higher QED values, LigGen
consistently produced ligands with good <strong>Lipinski compliance</strong>,
suggesting its balance between drug-likeness and innovation.
</Typography>
</Box>
{/* Alzheimer's Protein Analysis */}
<Box sx={{ marginBottom: 3 }}>
<Typography variant="h5" component="h2" gutterBottom>
Alzheimer’s-Related Protein Analysis
</Typography>
<Typography variant="body1" paragraph>
In the Alzheimer’s-related protein analysis, ligands generated by LigGen
consistently showed better binding affinities compared to those from
LigBuilder and Pocket2Mol. LigGen demonstrated a balanced approach
between ligand synthesizability and structural complexity, as seen in
its ability to produce ligands with moderate synthesizability scores.
</Typography>
</Box>
{/* Conclusion */}
<Box sx={{ marginBottom: 3 }}>
<Typography variant="h5" component="h2" gutterBottom>
Conclusion
</Typography>
<Typography variant="body1" paragraph>
Overall, the results indicate that LigGen is a robust tool for fragment-based
ligand generation, capable of producing high-quality ligands with
competitive docking scores and favorable synthesizability. LigGen provides
a novel method for generating ligands using a fragment-based de novo
drug design approach.
</Typography>
</Box>
</Container>
);
const otherToolsUI = () => (
<Container>
<Typography variant="h4" gutterBottom>
Other Tools
</Typography>
<Typography variant="body1">Explore additional functionalities and tools here...</Typography>
</Container>
);
const computationUi = () => {
return (
<Container maxWidth="md">
{submitJobUI()}
<Container>
<Box sx={{ mt: 4 }}>
<Typography variant="subtitle1">Job ID: {jobId}</Typography>
</Box>
</Container>
{checkStatusUI()}
<Typography variant="subtitle2">Status: {status}</Typography>
<Box>
{ligandDetailsDict && Object.keys(ligandDetailsDict).length > 0 ? (
Object.keys(ligandDetailsDict)
.map((key) => Number(key)) // Convert keys to numbers for numeric sorting
.sort((a, b) => a - b) // Sort numerically in ascending order
.map((key) => {
const ligandDetails = ligandDetailsDict[key.toString()]; // Access value using sorted key
return ligandDetailsUI(ligandDetails, key); // Pass key if needed
})
) : (
<Typography variant="h6" align="center" sx={{ mt: 4 }}>
No Ligand Details Available
</Typography>
)}
</Box>
</Container >
);
}
const ourResultsUI = () => (
<Container>
<Typography variant="h4" gutterBottom>
Results: CrossDock Dataset
</Typography>
<Typography variant="body1" gutterBottom>
We evaluated LigGen using the test set of the CrossDock dataset. The test set contains 100 diverse protein-ligand pairs. This table summarizes the molecular properties of the ligands generated for these targets. We compared LigGen against multiple baselines, including 3D-SBDD, Pocket2Mol, GraphBP, TargetDiff, DecompDiff, DiffSBDD, FLAG, DrugGPS, Lingo3DMol, and Frag2Seq. Baseline results are sourced from the paper “Fragment and Geometry Aware Tokenization of Molecules for Structure-Based Drug Design Using Language Models”.
We use the same evaluation matrix which was used in baselines. Vina Score estimates the binding affinity between generated molecules and given protein pockets; QED is a measure used to assess the drug-likeness of a molecule based on its molecular properties; SA estimates how easy it would be to synthesize a given chemical compound; Lipinski measures how well a molecule satisfies the Lipinski's rule of five , which evaluates the drug-likeness of a molecule; Diversity measures the average pairwise diversity (calculated as 1-Tanimotosimilarity) of generated molecules for a binding pocket; Time is the average time cost to generate 100 molecules for a protein pocket in the test set. All the Vina scores are calculated by QVina, and the chemical properties are calculated by RDKit .
</Typography>
<TableContainer component={Paper}>
<Table>
<TableHead>
<TableRow>
<TableCell><strong>Methods</strong></TableCell>
<TableCell><strong>Vina Score (↓)</strong></TableCell>
<TableCell><strong>QED (↑)</strong></TableCell>
<TableCell><strong>SA (↑)</strong></TableCell>
<TableCell><strong>Lipinski (↑)</strong></TableCell>
<TableCell><strong>Diversity (↑)</strong></TableCell>
<TableCell><strong>Time (s, ↓)</strong></TableCell>
</TableRow>
</TableHead>
<TableBody>
{[
["Test set*", "-6.87±2.32", "0.47±0.20", "0.72±0.14", "4.34±1.14", "-", "-"],
["3D-SBDD*", "-5.88±1.91", "0.50±0.17", "0.67±0.14", "4.78±0.51", "0.74±0.09", "15986.4±9851.0"],
["Pocket2Mol*", "-7.05±2.80", "0.57±0.16", "0.75±0.12", "4.93±0.27", "0.73±0.15", "2827.3±1456.8"],
["GraphBP*", "-4.71±4.03", "0.50±0.12", "0.30±0.09", "4.88±0.37", "0.84±0.01", "1162.8±438.5"],
["TargetDiff*", "-7.31±2.47", "0.48±0.20", "0.58±0.13", "4.59±0.83", "0.71±0.09", "~3428"],
["DecompDiff*", "-6.60±2.11", "0.49±0.21", "0.65±0.14", "4.49±1.02", "0.72±0.10", "~6189"],
["DiffSBDD*", "-7.17±3.28", "0.55±0.20", "0.72±0.12", "4.74±0.59", "0.71±0.07", "629.9±277.2"],
["FLAG*", "-6.38±3.24", "0.48±0.19", "0.70±0.15", "4.65±0.74", "0.70±0.14", "1289.1±378.0"],
["DrugGPS*", "-6.60±2.38", "0.46±0.21", "0.62±0.15", "4.49±0.99", "0.73±0.10", "1007.8±554.1"],
["Lingo3DMol*", "-7.25±1.69", "0.26±0.15", "0.65±0.08", "3.12±1.25", "0.48±0.12", "1481.9±1512.8"],
["Frag2Seq*", "-7.36±1.96", "0.64±0.15", "0.64±0.11", "4.98±0.11", "0.71±0.07", "48.8±14.6"],
["LigGen", "-7.50±1.83", "0.43±0.22", "0.68±0.10", "4.60±0.73", "0.69±0.13", "1271.335±3107.77"],
].map((row, index) => (
<TableRow key={index}>
{row.map((cell, cellIndex) => (
<TableCell key={cellIndex}>{cell}</TableCell>
))}
</TableRow>
))}
</TableBody>
</Table>
</TableContainer>
<br /> <br /><br /> <br /> <br />
<Typography variant="h4" gutterBottom>
Alzheimer’s-Related Protein Analysis
</Typography>
<Typography>
In this section, we compare the performance of LigGen by generating ligands for Alzheimer's-Related Targets Beta-secretase 1, Monoamine Oxidase B and Acetylcholinesterase against tools LigBuilder V3 and Pocket2Mol. LigBuilder V3 and Pocket2Mol are popular ligand-building tools based on Genetic Algorithms and Deep Learning, respectively. We have generated results for three different targets with PDB IDs 2G94, 2V5Z, and 7D9O. The designed ligands were evaluated based on their Binding Affinities (using AutoDock Vina).
A total of 100 ligands were generated for each protein using all three programs, LigGen, Pocket2Mol and LigBuilder. Subsequently, these compounds underwent docking using AutoDock-Vina.
</Typography>
<div style={{ display: "flex", justifyContent: "center", gap: "20px", alignItems: "center" }}>
<div>
<img src="/images/ba_7d9o.png" alt="7D9O: Docking Scores" style={{ width: "350px", height: "auto" }} />
</div>
<div>
<img src="/images/ba_2v5z.png" alt="2V5Z: Docking Scores" style={{ width: "350px", height: "auto" }} />
</div>
<div>
<img src="/images/ba_2g94.png" alt="2G94: Docking Scores" style={{ width: "350px", height: "auto" }} />
</div>
</div>
</Container>
);
return (
<Box>
<AppBar position="static">
<Toolbar>
{/* Logo or Title */}
<Box display="flex" alignItems="center" sx={{ marginRight: 2 }}>
<Typography variant="h6" noWrap component="div">
LigGen Webserver
</Typography>
</Box>
<Tabs value={activeTab} onChange={handleTabChange} indicatorColor="secondary"
textColor="inherit"
centered sx={{ flexGrow: 0.5 }} >
<Tab label="Introduction" />
<Tab label="Computation" />
<Tab label="Our Results" />
<Tab label="Other Tools" />
</Tabs>
</Toolbar>
</AppBar>
<Box sx={{ p: 3 }}>
{activeTab === 0 && introductionUI()}
{activeTab === 1 && computationUi()}
{activeTab === 2 && ourResultsUI()}
{activeTab === 3 && OtherTools()}
</Box>
</Box>
);
};
export default App;

App.js at commit 4563b04, no license · at the source

Overview

Authors: Anshul Yadav¹, Natarajan Arul Murugan¹

Department of Computational Biology, Indraprastha Institute of Information Technology, Okhla Phase 3, New Delhi, 110020 India

Institutions: Indraprastha Institute of Information Technology Delhi (India)

Journal: Scientific reports, volume 16, issue 1, article 20477

Dates: received 21 December 2025; accepted 7 April 2026; published online 2 May 2026

Type: Research article · Language: English

License: CC BY-NC-ND

Identifiers: DOI 10.1038/s41598-026-48239-2 · PMID 42069853 · PMCID PMC13328569 · OpenAlex W7159971684

Open access: gold, a free copy (OpenAlex)

Status: code verified

Categories: computational modeling (no new data) (modality), human (organism), computational (subfield)

Methods: Machine learning

Keywords: Fragment-based drug design, Recurrent neural networks, Monte Carlo simulated annealing, Chemistry, Computational biology and bioinformatics, Drug discovery

MeSH: Drug Design*, Acetylcholinesterase, Amyloid Precursor Protein Secretases, Aspartic Acid Endopeptidases, Drug Discovery, Humans, Ligands, Monoamine Oxidase, Monte Carlo Method, Recurrent Neural Networks (* major topic)

Topic: Computational Drug Discovery Methods (Computational Theory and Mathematics, Computer Science), according to OpenAlex

Funding: SERB-SURE (SUR/2022/005296)

Citations: not cited yet (Europe PMC); 59 references in the paper

Abstract

The abstract is not reproduced here: the paper's license (CC BY-NC-ND) does not allow it. Read it in the paper, at the publisher or on Europe PMC.

Repository

Its files are read in the Code ↔ Paper reader above, with 12 matches between paragraphs and lines of code.

neurocare-iiitd/LigGen

License: none: the authors keep all their rights

State: the link answers, verified on 28 September 2026

Evidence: files inventoried

Commit: 4563b04ddd71739d4d555680d3bdbbf6761b5825, 15 June 2026

Languages: Python (9), JavaScript (7), Jupyter (2)

Size: 76 files, 18 scripts

Software Heritage: not archived

Found in: “Data availability”

Holds: README, environment (environment.yml), tests, 1 notebook

Not found: license file, CITATION.cff, continuous integration, documentation

Tools: PyTorch (5 files), RDKit (5 files), NumPy (2 files), Matplotlib (1 file), SciPy (1 file)

Availability: 1 check, the latest on 28 September 2026: the link answers

28 September 2026: the link answers

19 files

.ipynb_checkpoints/main-checkpoint.ipynb, Jupyter, 206 lines
generate_fragments.py, Python, 68 lines
generate_ligands.py, Python, 390 lines, 1 match
mcsa.py, Python, 449 lines, 2 matches
rnn_config.py, Python, 18 lines
rnn_selfies.py, Python, 65 lines, 2 matches
selfies_dataset.py, Python, 51 lines
server/frontend/src/App.js, JavaScript, 766 lines, 4 matches
server/frontend/src/App.test.js, JavaScript, 8 lines
server/frontend/src/ExampleFile.js, JavaScript, 2,501 lines
server/frontend/src/index.js, JavaScript, 17 lines
server/frontend/src/otherTools.js, JavaScript, 68 lines, 1 match
server/frontend/src/reportWebVitals.js, JavaScript, 13 lines
server/frontend/src/setupTests.js, JavaScript, 5 lines
server/server.py, Python, 143 lines
tools.ipynb, Jupyter, 30 lines
train_fragment_generator.py, Python, 209 lines, 2 matches
utils.py, Python, 254 lines
README.md, Text, 121 lines

At the source: github.com/neurocare-iiitd/LigGen

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
18 scripts, each with its path and the digest of its content;
12 matches between paragraphs of the paper and lines of the code (method lexical-v1);
neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

Datasets cited

rcsb.org/structure/2g94, at PDB; found in the references
rcsb.org/structure/2v5z, at PDB; found in the references
rcsb.org/structure/7d9o, at PDB; found in the references
zenodo:17733479, at Zenodo; found in the references

Code and data availability statement

The paper has a code and data availability statement. Its license (CC BY-NC-ND) does not allow reproducing it here; in short, from what the harvester recognized in it:

it points to the authors' code: neurocare-iiitd/LigGen

Read it in the paper: doi.org/10.1038/s41598-026-48239-2.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 28 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 2 authors, 6 keywords, 10 MeSH terms, 1 funder, 41 references.

Cite

This paper

Yadav, A., & Murugan, N. A. (2026). LigGen-a GEN-AI based ligand generation approach for de-novo drug design. Scientific reports, 16(1), 20477. https://doi.org/10.1038/s41598-026-48239-2

BibTeX

@article{yadav2026liggen, author = {Yadav, Anshul and Murugan, Natarajan Arul}, title = {{LigGen-a GEN-AI based ligand generation approach for de-novo drug design}}, journal = {Scientific reports}, year = {2026}, month = may, volume = {16}, number = {1}, pages = {20477}, publisher = {Nature Publishing Group}, issn = {2045-2322}, doi = {10.1038/s41598-026-48239-2}, url = {https://doi.org/10.1038/s41598-026-48239-2}, pmid = {42069853}, pmcid = {PMC13328569} }

RIS

TY - JOUR AU - Yadav, Anshul AU - Murugan, Natarajan Arul TI - LigGen-a GEN-AI based ligand generation approach for de-novo drug design T2 - Scientific reports J2 - Sci Rep PY - 2026 DA - 2026/05/02 VL - 16 IS - 1 SP - 20477 SN - 2045-2322 PB - Nature Publishing Group DO - 10.1038/s41598-026-48239-2 UR - https://doi.org/10.1038/s41598-026-48239-2 LA - en ER -

CSL-JSON

{ "id": "10.1038/s41598-026-48239-2", "type": "article-journal", "title": "LigGen-a GEN-AI based ligand generation approach for de-novo drug design", "container-title": "Scientific reports", "author": [ { "family": "Yadav", "given": "Anshul" }, { "family": "Murugan", "given": "Natarajan Arul" } ], "container-title-short": "Sci Rep", "volume": "16", "issue": "1", "page": "20477", "DOI": "10.1038/s41598-026-48239-2", "PMID": "42069853", "PMCID": "PMC13328569", "ISSN": "2045-2322", "publisher": "Nature Publishing Group", "URL": "https://doi.org/10.1038/s41598-026-48239-2", "language": "en", "issued": { "date-parts": [ [ 2026, 5, 2 ] ] } }

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1038/s41598-026-53415-5 [code]: Computational design and immunoinformatics validation of a T cell multi-epitope vaccine targeting glioblastoma stem cells.
Journal: Scientific reports
In common: RDKit, PyTorch, SciPy, 2 other tools, computational modeling (no new data), 2 references
[2] doi:10.1021/acs.jmedchem.6c00231 [code]: Applying Deep-Learning-Driven <i>De Novo</i> Design to Hit Identification: A Case Study on A<sub>2A</sub> Adenosine Receptor Antagonists.
Journal: Journal of medicinal chemistry
In common: RDKit, PyTorch, SciPy, 1 other tool, 2 references
[3] doi:10.1021/acs.jcim.6c01299 [code]: Targeting BCL-2 through Deep Learning-Based Drug Repurposing: A Multimodal Approach Combining Diffusion-Based Generative Modeling, Neural Relational Inference, and In Vitro Validation.
Journal: Journal of chemical information and modeling
In common: RDKit, PyTorch, SciPy, 2 other tools, computational modeling (no new data)
[4] doi:10.1093/bib/bbag118 [code]: Drug screening for α-synuclein aggregation inhibitors via multimodal graph neural network.
Journal: Briefings in bioinformatics
In common: RDKit, PyTorch, SciPy, 2 other tools, computational modeling (no new data)
[5] doi:10.1038/s41586-026-10670-w [code]: Zero-shot design of drug-binding proteins via neural iterative selection-expansion.
Journal: Nature
In common: RDKit, PyTorch, SciPy, 2 other tools, computational modeling (no new data)
[6] doi:10.1038/s41467-026-75444-4 [code]: Structural insights enable drug discovery for the neuronal NBCn2 carbonate transporter.
Journal: Nature communications
In common: RDKit, SciPy, Matplotlib, 1 other tool, 1 reference
[7] doi:10.1098/rstb.2024.0461 [code]: Shallow recurrent decoders for neural and behavioural dynamics.
Journal: Philosophical transactions of the Royal Society of London. Series B, Biological sciences
In common: PyTorch, SciPy, Matplotlib, 1 other tool, computational, 1 reference
[8] doi:10.1038/s42003-026-10957-8 [code]: Brain defence by the extracellular matrix protein Cochlin.
Journal: Communications biology
In common: RDKit, PyTorch, SciPy, 2 other tools
[9] doi:10.1371/journal.pone.0345854 [code]: Shedding light on neural learning to rank models for anticancer drug prioritization.
Journal: PloS one
In common: RDKit, PyTorch, SciPy, 2 other tools
[10] doi:10.1093/nar/gkag706 [code]: scDifformer: diffusion-based post-training for virtual cell modeling across large-scale single-cell data.
Journal: Nucleic acids research
In common: RDKit, PyTorch, SciPy, 2 other tools

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.