Biological Database Search

Overview

The command bar connects to 8 major biological databases through a unified search interface. Select a tag (e.g., Protein, Gene, Structure) then type your query to search any database without leaving the canvas.

Each search result creates a canvas node tailored to its data type — 3D protein structures rendered in Mol*, gene information cards, literature citations, interactive KEGG pathway maps, and circular genome visualizations. Results are cached locally for fast repeat access and can be arranged, connected, and annotated directly on the canvas.

Quick Reference

UniProt (Protein Search)

UniProt searches query the reviewed Swiss-Prot database, returning curated protein entries with function annotations, domain architecture, disease associations, and cross-references. When a protein has an associated PDB structure, the result node renders a 3D viewer. Otherwise it displays a detailed InfoCard.

Example Queries

Tips

• Gene symbols like TP53 or EGFR work well for human proteins. For other organisms, append the species: TP53 mouse.
• Partial protein names use fuzzy matching — tumor suppressor p53 finds the same entry as TP53.
• Results include GO annotations, subcellular location, and links to external databases (PDB, InterPro, Pfam).

NCBI Gene

NCBI Gene searches return gene metadata including chromosomal location, aliases, RefSeq identifiers, and functional summaries. Results appear as Gene InfoCards with direct links to the NCBI Gene page.

Example Queries

Tips

• Use HGNC-approved gene symbols for the most reliable results. Aliases (e.g., HER2 for ERBB2) are resolved but may return multiple matches.
• The search defaults to human genes. Specify the organism explicitly if you need a different species: BRCA1 rat.
• Gene IDs (numeric) perform direct lookups and are the fastest query type.

RCSB PDB (Structure Search)

PDB searches query the RCSB Protein Data Bank for experimentally determined 3D structures. Results create a MolstarNode — a fully interactive molecular viewer powered by Mol* with rotation, zoom, surface/cartoon toggles, and chain highlighting.

Example Queries

Tips

• PDB IDs are exactly 4 characters (one digit followed by three alphanumeric characters, e.g., 1CRN, 6LU7). When you enter a valid PDB ID, the structure loads directly without a search step.
• Add an experimental method to narrow results: kinase X-ray, antibody cryo-EM, peptide NMR.
• The Mol* viewer supports multiple representations (cartoon, ball-and-stick, surface) and can highlight individual chains, ligands, or residue ranges.

PubMed (Literature Search)

PubMed searches query titles, abstracts, and MeSH (Medical Subject Headings) terms across the full MEDLINE database. Results appear as Citation InfoCards showing the title, authors, journal, year, and abstract excerpt.

Example Queries

Tips

• Boolean operators (AND, OR, NOT) work as expected. Use them to refine broad topics.
• MeSH terms improve precision for established concepts. PubMed automatically maps common terms to their MeSH equivalents.
• Prefix a PMID with PMID: for direct paper lookup without a search round-trip.
• Results are sorted by relevance. The most recent and highest-cited papers appear first.

ChEMBL (Compound Search)

ChEMBL searches query the EMBL-EBI database of bioactive molecules with drug-like properties. Results include molecular structure, clinical development phase, mechanism of action, and target information.

Example Queries

Tips

• Both common drug names and ChEMBL accession IDs work. Accession IDs (e.g., CHEMBL25) perform direct lookups.
• Compound InfoCards display the clinical phase (Approved, Phase I-III, Preclinical), molecular formula, and key physicochemical properties.
• Target information links compounds to their protein targets, enabling cross-referencing with UniProt results on the same canvas.

NCBI GEO (Dataset Search)

GEO searches query the Gene Expression Omnibus for publicly available gene expression, methylation, and other functional genomics datasets. Results appear as Dataset InfoCards showing the GSE accession, title, organism, sample count, and platform.

Example Queries

Tips

• Be specific about organism, method, and tissue type. scRNA PBMC human returns more relevant results than single cell.
• GSE accession numbers (e.g., GSE198765) perform direct lookups and are the fastest way to retrieve a known dataset.
• Dataset InfoCards show the number of samples and platform, helping you assess dataset suitability before downloading.

KEGG (Pathway Search)

KEGG searches query the Kyoto Encyclopedia of Genes and Genomes for metabolic pathways, signaling cascades, and disease pathways. Results create a PathwayNode — an interactive, zoomable pathway map with clickable entities (genes, compounds, reactions).

Example Queries

Tips

• Use standard pathway names as they appear in KEGG. Common abbreviations (MAPK, mTOR, TCA cycle) are recognized.
• KEGG pathway IDs (e.g., hsa04110) perform direct lookups. The hsa prefix indicates human pathways.
• Double-click any entity (gene box, compound circle) in the pathway map to view its details or search for it in the corresponding database.
• Pathway maps are interactive: zoom with scroll, pan by dragging, and click entities to highlight connected reactions.

NCBI Nucleotide (Sequence Search)

NCBI Nucleotide searches query the GenBank and RefSeq sequence databases for genomes, plasmids, chromosomes, and individual gene sequences. Results create a CgviewNode — an interactive circular or linear genome map rendered with Cgview.js, showing annotated features like genes, regulatory elements, and restriction sites.

Example Queries

Tips

• Well-annotated sequences (reference genomes, common plasmids) produce the richest genome maps with full feature annotations.
• RefSeq accession numbers (e.g., NC_000913) and GenBank accessions perform direct lookups.
• Circular maps are generated for circular molecules (plasmids, bacterial chromosomes). Linear maps are used for linear genomes and chromosomal segments.
• Zoom into specific regions of the genome map to see individual gene annotations, reading frames, and regulatory features.

Search Behavior

Caching

Search results are cached locally to speed up repeat queries and reduce load on external databases.

Rate Limits

Error Handling

• If an external database is temporarily unavailable, the search returns "No results found" rather than an error. This is graceful degradation — retry after a few seconds.
• Each database query has a timeout of 3-5 seconds. Slow upstream responses are terminated cleanly.
• No 500 errors propagate to the UI. All failures are caught and displayed as user-friendly messages.