Large candidate space
Plant genomes often encode large Family-1 UGT repertoires, making exhaustive experimental screening expensive.
From glycosylation hypothesis to experimentally prioritized UGT candidates
A candidate-centred research environment integrating evolutionary context, donor-aware structural evidence, biological context and interpretable multi-evidence prioritization for plant Family-1 UDP-glycosyltransferases.
UGTLD organizes heterogeneous evidence around each candidate and preserves its origin through prioritization and decision.
Plant genomes often encode large Family-1 UGT repertoires, making exhaustive experimental screening expensive.
Evolutionary, structural, biological and reaction-related evidence are commonly analyzed in separate tools.
Missing or unavailable evidence must remain explicit—not become artificial confidence, zero, or a worst rank.
System priority supports—not replaces—the researcher’s experimental decision.
Every axis remains inspectable at candidate level. Reaction Geometry is diagnostic and never participates in RRF.
Task-specific functional anchors, candidate placement and candidate-to-anchor evolutionary context.
Candidate structure, donor-conditioned sequential docking, acceptor context and reaction-site evidence.
Dataset-supported expression, omics and functional context—only when scientifically available.
Unavailable → omitted, never zeroReactive atom mapping, distance, angle and geometric plausibility for mechanistic inspection.
Optional deeper structural and AI-supported evidence for a frozen or designated subset, not the initial ranking foundation.
Available ranking evidence is integrated with a missingness-aware strategy. Unavailable evidence is omitted—not scored as zero, assigned a worst rank, or replaced by candidate order.
Each candidate retains axis-level evidence and its scientific state.
Unavailable evidence remains unavailable and is not converted to zero.
Results remain connected to jobs, artifacts, protocol and analysis context.
System prioritization and experimental decisions remain separate.
These views use the same persisted project, candidate, evidence and output lineage exposed by the research workspace.
The same scientific assets feed workspace views, reports, exports and publication-quality charts.
Define the biological question, establish a candidate set, inspect evidence, compare candidates and preserve the complete research lineage.
Multi-Species UGT Gene Mining & Molecular Docking Platform
Upload FASTA file or paste sequences for PSPG motif heuristic scanning and UGT family classification.
This module uses heuristic rules to scan for plant secondary product glycosyltransferase (PSPG) motifs in protein sequences. The detection checks key conserved positions within the C-terminal PSPG domain (~44 aa).
Open a project to continue its task, candidate set, evidence and delivery lineage.
| Latin Name | Chinese | Group | UGT Count | PSPG | Avg Length | Action |
|---|---|---|---|---|---|---|
| Loading... | ||||||
Import a curated UGT FASTA directly into this project’s UGT Table, or upload a full proteome for separate PF00201 identification. Uploading a proteome alone does not identify UGTs.
Rows come from the selected project resource. Explicit header annotations are shown as supplied; missing family labels remain unannotated.
| Sequence QC | Source | Action | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Import or select a project FASTA resource to view records. | ||||||||||
Protein-level QC, PSPG and conservative reference context. Reference functions are not candidate substrate predictions.
| Original ID | Label | Length | PSPG | QC | Family | Subfamily-like | Closest named | Closest characterized | Phylogenetic placement | Confidence | Structure | Action | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Load a species to begin. | |||||||||||||
UGT requires both sugar donor and substrate binding for glycosylation. Two-step docking protocol:
Step 1 UGT + Sugar Donor → Complex →
Step 2 UGT·Donor + Substrate → Ternary Complex
| Gene ID | Donor | Step1 Score UGT+Donor | Substrate | Step2 Score Complex+Sub | Combined Score | Catalytic Pose | Evidence |
|---|---|---|---|---|---|---|---|
| Submit a docking job first. | |||||||
Open a project to review its research question.
No project analysis loaded.
No project analysis loaded.
No project analysis loaded.
Open a project to view prioritization results.
| Rank | Candidate | Priority | Evidence | Evolution | Structural | Biological | Decision | |
|---|---|---|---|---|---|---|---|---|
| Open a project to view results. | ||||||||
Observe, select, compare, prioritize, and export candidates without turning evidence coverage into a score.
| Select | Candidate | Decision | Priority | System priority | Evolution | Structure | Reverse screening | Geometry | Boltz | Biology | Coverage | Next action |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Load a project to inspect candidates. | ||||||||||||
Select a candidate from Candidate Center to inspect evidence, provenance, and report-ready outputs.
Dataset-supported expression and pathway coupling for the current project.
| Candidate | Expression readiness | Target pathway | Spearman rho | Biological rank | Provenance |
|---|---|---|---|---|---|
| No biological evidence loaded. | |||||
Search UGT-related literature with PubMed, summarize with DeepSeek AI.
AI summarization of search results. Requires DeepSeek API Key (configure in Settings).
Submit structure prediction, docking, and affinity analysis tasks. AI backends (Boltz-2 / AlphaFold3 / RoseTTAFold2) integrate via API.
| Task ID | Type | Targets | Substrates | Status | Created | Result |
|---|---|---|---|---|---|---|
| No AI tasks submitted yet | ||||||
Simulate transcriptional effects of silencing (KO) or overexpressing (OE) target genes using Ridge regression on local co-expression networks.
Structure-informed mutation planning integrating catalytic context, ligand contacts, pocket/channel annotations, conservation and stability evidence.
Context-aware structural evidence for the selected candidate.
3Dmol.js-powered protein structure visualization. Load PDB files, highlight mutations, import pocket/channel data, and export figures.
RoseTTAFold2-powered protein 3D structure prediction. Upload single sequence or batch-predict from Candidate Pool.
Paste a single FASTA sequence or upload a .fasta file. RoseTTAFold2 will generate a 3D structure model.
Select genes from Candidate Pool for batch structure prediction. All results will be linked back to candidate scores.
| Gene ID | Species | Status | pLDDT | Method | Created | Action |
|---|---|---|---|---|---|---|
| Submit a prediction job to see results. | ||||||
Live Confirmatory26 chart manifest, SVG exports, TSV source tables, and publication-ready assets.
Filter durable jobs by status, tool, project, or candidate. Formal queues are read-only here: no delete, reset, rerun-all, or recreate controls.
| Job ID | Type | Status | Progress | Project | Candidate | Created | Duration | Action |
|---|---|---|---|---|---|---|---|---|
| No jobs yet. Submit a docking or mining job to see results here. | ||||||||
Live read-only status for UGTLD R2.1 Confirmatory26. No reset, delete, rerun, recreate, or Gold access controls are exposed here.
Enter your DeepSeek API key manually. The key is stored only on the backend server — never in browser localStorage, logs, or URL parameters.
Tools required for custom ligand conversion (SMILES to 3D, format conversion).
Live status of the backend toolchain (structure prediction, docking, ligand conversion, parametrization). States are probed at runtime and cached 60s server-side.
The final traceable research output: context, candidates, evidence, integrated priority, decisions, charts, report, and provenance.