Available Databases
SmartsBio connects to 89+ major public bioinformatics databases across sequences, structures, variants, pathways, clinical data, literature, and patents. Each database is accessed through one or more tools in the API.
How it works: Databases are accessed via tools — each tool wraps one or more database APIs. Call them directly with
POST /v1/tools/:id/run or let the agent automatically pick the right databases when you use the Query endpoint.Database Reference
89 databases organized by data type. Click a database name to read the full parameter reference.
Drug Discovery3 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| ADMET Property Predictor | admet-predictor | — | Predict ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) properties for one or more small molecules using RDKit descriptors and DeepChem pre-trained models. RDKit-computed properties (fast, no model): MW, LogP, TPSA, HBA, HBD, RotBonds, Lipinski Ro5 compliance, Veber compliance. DeepChem model predictions: aqueous solubility (LogS, ESOL model), Caco-2 permeability, human intestinal absorption (HIA), blood-brain barrier penetration (BBB), CYP3A4/2D6/2C9 inhibition flags, hERG cardiotoxicity flag. Molecules are scored with a composite ADMET score (0–1) and ranked. Accepts a SMILES string or a file from molecule-variant-generator output. Returns ranked JSON + CSV. topN limits output to top N molecules by composite score. Runs on ECS Fargate (CPU). Runtime: 5–20 minutes depending on library size. | smiles, inputFile, topN |
| Molecule Variant Generator | molecule-variant-generator | — | Generate structural variants (analogs) of a query small molecule using RDKit. Applies three complementary strategies: R-group enumeration (replaces substituents with common drug-like groups), bioisostere replacement (classical medicinal chemistry transformations: carboxylic acid → tetrazole, benzene → pyridine, etc.), and fragment growing (extends the scaffold at available attachment points). All variants are automatically filtered by Lipinski Rule of Five (MW ≤ 500, LogP ≤ 5, HBA ≤ 10, HBD ≤ 5) and Veber rules (TPSA ≤ 140, RotBonds ≤ 10) to keep only drug-like candidates. Output: JSON file with up to numVariants molecules, each with SMILES + computed properties (MW, LogP, TPSA, HBA, HBD, RotBonds). Runs on ECS Fargate (CPU). Runtime: 5–15 minutes. Use as the first step of a lead optimization workflow before ADMET prediction and binding affinity ranking. | smiles, numVariants, optimizationGoals |
| OpenFDA Drug Database Tools | openfda-drug-toolkit | — | Comprehensive drug information from FDA, EMA, and WHO databases including approvals, adverse events, labeling, and recalls | tool_type, query, limit, patient_reaction, seriousness |
Structural Biology1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Molecular Docking | molecular-docking | — | Blind molecular docking of small molecules onto protein targets using DiffDock-L (MIT, MIT/Stanford). Diffusion-based approach: automatically discovers binding pockets without prior knowledge. Success rate: 43% (RMSD <2Å), DockGen generalization benchmark: 22.6% — state-of-the-art blind docking. Generates multiple ranked docking pose candidates with confidence scores. Complements Chai-1 (co-folding from scratch); DiffDock-L docks onto an existing structure. Runs on GPU-accelerated AWS Batch (Spot g5.xlarge). Use for drug discovery: identify binding mode and rank ligand candidates. | proteinPDB, ligandSMILES, numPoses, inferenceSteps |
Proteomics12 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Molecular Dynamics Simulation | molecular-dynamics | — | Run molecular dynamics (MD) simulation to validate protein stability using OpenMM (MIT/LGPL, Stanford). Simulates the protein in explicit water + ions at physiological conditions (300 K, 1 atm, 150 mM NaCl). Validates dynamic stability — ThermoMPNN checks static ΔΔG; OpenMM checks real behaviour over nanoseconds. OpenMM 8 supports ML potentials alongside physics-based force fields. Outputs: trajectory file, RMSD stability plot, per-residue flexibility (RMSF), and average structure PDB. Runtime: ~100–300 ns/day on g5.xlarge (A10G); 50–100 ns sufficient for stability validation. Runs on GPU-accelerated AWS Batch (Spot g5.xlarge). Cost: ~$3–10 per job. | proteinPDB, simulationLength, forcefield, temperature, solvent |
| Protein Complex Prediction | protein-complex-prediction | — | Predict the 3D structure of multi-molecule complexes using Chai-1 (Apache 2.0, ChaiDiscovery). Co-folds proteins, small molecules, nucleic acids, glycans, and ions simultaneously in one model. Protein-ligand: 77% success rate (vs AlphaFold3 76%). Protein-protein multimers: 75.1% DockQ (beats AlphaFold-Multimer 67.7%). Antibody-antigen: 47.9% DockQ in single-sequence mode. Runs on GPU-accelerated AWS Batch (Spot g5.xlarge). Use when you need multi-molecule complex structure prediction without an AlphaFold3 licence. | sequences, ligandSMILES, useTemplates, numRecycles |
| Protein Sequence Co-design | protein-sequence-codesign | — | Generate protein sequence and 3D structure simultaneously using DPLM-2 diffusion model. Performs true joint sequence+structure co-design in a single diffusion pass — no separate inverse folding step. Supports unconditional generation and motif-constrained scaffolding (design a protein around a fixed active site). 650M DPLM-2 model outperforms 3B-scale baselines (ICLR 2025 / ICML 2025 Spotlight). Runs on GPU-accelerated AWS Batch (Spot g5.xlarge). Use for de novo protein generation or motif-anchored scaffold design. | mode, proteinLength, numSamples, temperature, motifPDB |
| Protein Sequence Generation | protein-sequence-generation | — | Generate diverse protein sequences within a protein family using ProGen2 (Salesforce, BSD-3). Trained on 1 billion protein sequences, ProGen2 produces natural-looking sequences that respect the evolutionary grammar of the target protein family. Provide a family name (e.g. "GFP", "serine protease") or a seed sequence, and the model generates diverse variants with perplexity scores for quality ranking. Runs on GPU-accelerated AWS Batch (Spot g5.xlarge). Ideal as a fast pre-screening step before expensive GPU design runs. | proteinFamily, numSequences, temperature, modelSize, maxLength |
| Protein Sequence Infilling | protein-sequence-infilling | — | Design protein scaffolds that preserve a required structural motif using DPLM-2 inverse folding. Given a PDB structure with a fixed functional site (e.g. catalytic triad, disulfide bonds, binding loop), generates diverse scaffold sequences that maintain the exact geometry of the fixed residues. Useful for enzyme engineering and epitope grafting where the active site geometry is non-negotiable. Runs on GPU-accelerated AWS Batch (Spot g5.xlarge). | motifPDB, motifResidues, scaffoldLength, numDesigns, confidenceThreshold |
| Protein Variant Scoring | protein-variant-scoring | — | Zero-shot fitness scoring of protein variants using ProGen2 log-likelihood. Ranks point mutations (e.g. A123G), insertions, deletions, and combinatorial variants by their log-likelihood delta relative to the wildtype — no experimental training data required. Handles both substitutions AND indels, unlike most tools that only score substitutions. Ideal for pre-screening thousands of variants cheaply before committing to wet lab synthesis. Runs on GPU-accelerated AWS Batch (Spot g5.xlarge). BSD-3 license (Salesforce). | baseSequence, mutations, scanAllSingle, modelSize |
| Antibody Humanization | antibody-humanization | — | Humanize antibody/nanobody sequences and score immunogenicity using BioPhi (MIT, Merck). Two complementary methods: Sapiens — deep learning CDR humanization while preserving binding specificity. OASis — 9-mer humanness scoring against Observed Antibody Space; correlates with clinical immunogenicity. Critical downstream step for all Boltz nanobody designs before therapeutic development. A nanobody with excellent binding but poor OASis humanness score will fail in clinical trials. Runs on ECS Fargate (CPU). Fast: 1–2 minutes per antibody. | antibodySequence, humanizationMethod, species, preserveBinding |
| Protein Function Annotation | protein-function-annotation | — | Predict protein function (GO terms) and enzyme class (EC number) using DeepFRI (BSD-3, Flatiron Institute). GCN on contact maps + protein language model features for accurate zero-shot function prediction. Predicts: Molecular Function (MF), Biological Process (BP), Cellular Component (CC), and EC class. Identifies functional residues — which specific amino acids drive each predicted GO term. Critical quality control: ensures designed proteins have the intended functional annotation. Runs on ECS Fargate (CPU). Fast: 1–3 minutes per protein. | proteinInput, ontology, identifyFunctionalResidues |
| Protein Stability Prediction | protein-stability-prediction | — | Predict thermodynamic stability change (ΔΔG) for protein point mutations using ThermoMPNN (UNC). ThermoMPNN-D extends prediction to double mutants. Fast CPU inference: screens hundreds of variants in under a minute. ΔΔG < 0 kcal/mol = stabilising mutation; ΔΔG > 0 = destabilising. Critical gatekeeper: filter unstable designs before committing to expensive GPU runs or wet lab synthesis. License: CC BY-ND 4.0 (allows commercial use and self-hosting; prohibits modifying the model itself). Runs on ECS Fargate (CPU). Fast: <1 minute for single scan. | proteinPDB, mutations, scanMode, includeDoubles |
| smarts.bio BLAST (protein) | smarts-blast | — | FIRST CHOICE for any protein BLAST. A real NCBI BLAST+ blastp against UniProtKB/Swiss-Prot, running on our own infrastructure: answers in about one second, synchronously, in this conversation — no queue, no rate limit, no background process to poll. Returns percent identity, E-value, coverage, bit score and a UniProt accession per hit, so it gives the alignment statistics a reviewer would ask for. USE IT WHEN: the user asks to BLAST a protein sequence, wants to identify a protein, or wants homologs/orthologs with real alignment statistics. DO NOT USE IT FOR: nucleotide/DNA/RNA queries (it is protein-only — use ebi-job-dispatcher with tool_name 'ncbiblast' and a nucleotide database), or when the user specifically needs a database other than Swiss-Prot such as full UniProtKB/TrEMBL, PDB, nr or nt (use ebi-job-dispatcher, or ncbi-blast for NCBI-only databases). Swiss-Prot is curated and covers essentially every well-characterised protein, so it answers 'what is this protein?' for almost all real queries. Zero hits is a real, trustworthy answer meaning no significant Swiss-Prot match — report it as such rather than retrying elsewhere. | sequence, max_hits, evalue |
| STRING Database Tools | string-db | — | Comprehensive plugin for protein-protein interaction analysis using STRING database | tool_type, identifiers, species, required_score, limit |
| UniProt Database Tools | uniprot-toolkit | — | Comprehensive plugin for UniProt protein database operations including search, fetch, and ID mapping | tool_type, query, ids, organism, format |
Bowtie21 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Bowtie2 - Fast Read Alignment | bowtie2-toolkit | — | Ultrafast and memory-efficient tool for aligning sequencing reads to large reference genomes. Supports both end-to-end and local alignment modes with comprehensive parameter control for various sequencing platforms and read types. | command, referenceGenome, inputReads, outputFile, outputFormat |
Bwa1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| BWA - Burrows-Wheeler Aligner | bwa-toolkit | — | Fast and memory-efficient alignment tool for short DNA sequencing reads against large reference genomes using Burrows-Wheeler Transform. Supports multiple alignment algorithms and paired-end sequencing data with comprehensive parameter control. | command, referenceGenome, inputReads, outputFile, threads |
Hisat21 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| HISAT2 - Hierarchical Indexing for Spliced Alignment of Transcripts 2 | hisat2-toolkit | — | Fast and sensitive splice-aware alignment of RNA-seq reads to a reference genome using hierarchical graph-based indexing. | command, indexPrefix, referenceGenome, inputReads, outputFile |
3d Protein Structures1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| AlphaFold 3D Protein Structure Database [PRIMARY FOR 3D STRUCTURES] | alphafold-db | — | 🧬 PRIMARY TOOL FOR 3D PROTEIN STRUCTURE VISUALIZATION 🧬 AI-predicted 3D protein structure visualization and analysis. Access 214+ million protein structure predictions with confidence scores. ALWAYS USE THIS for 3D protein structures, molecular visualization, and structural analysis queries. | query_type, identifier, organism, confidence_threshold, format |
Preprints3 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| arXiv Search | arxiv-search | — | Search arXiv preprint repository for cutting-edge research and early-stage scientific papers | query, category, max_results, sort_by, sort_order |
| bioRxiv Search | biorxiv-search | — | Search bioRxiv preprint repository for cutting-edge biological research and early-stage life science papers | query, max_results, start_date, end_date, cursor |
| medRxiv Search | medrxiv-search | — | Search medRxiv preprint repository for cutting-edge health sciences research and early-stage medical papers | query, max_results, start_date, end_date, cursor |
Genomics4 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Biograph Knowledge Graph | biograph | — | Query the Biograph Knowledge Graph for genes, proteins, diseases, variants, pathways, and drugs with their relationships | query_type, entity_type, entity_id, search_term, limit |
| Ensembl | ensembl | — | Comprehensive plugin for Ensembl genome browser and annotation database - providing access to vertebrate genomic data. Includes COORDINATE LIFTOVER (toolType=map): convert genomic coordinates between assemblies, e.g. GRCh37/hg19 to GRCh38/hg38, using Ensembl's assembly-mapping service — no local chain files needed. Use this whenever coordinates from one genome build must be compared against data in another. | toolType, region, sourceAssembly, targetAssembly, species |
| Gene Ontology (GO) Toolkit | go-toolkit | — | Comprehensive toolkit for Gene Ontology analysis including term search, annotations, and enrichment | operation, query, go_id, identifiers, organism |
| NCBI-Gene | ncbi-gene | — | Search NCBI Gene database for gene information, locations, and annotations | query, organism, gene_type, chromosome, retmax |
Chembl1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| ChEMBL Database Tools | chembl-toolkit | — | Comprehensive plugin for ChEMBL bioactivity database operations including compound search, target analysis, and drug discovery data | tool_type, query, chembl_id, smiles, target_type |
Clingen1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| ClinGen - Gene-Disease Validity | clingen | — | ClinGen Gene Curation Expert Panel classifications of a gene's disease relationships: disease (MONDO), mode of inheritance (Autosomal dominant/recessive, X-linked), and validity classification (Definitive / Strong / Moderate / Limited / Disputed / Refuted). Use for the clinical-genetics rigor of the Disease Associations section — the formal gene→disease causality framework. | gene |
Clinical Medicine3 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Clinical Tables Search Service | clinical-tables-toolkit | — | Comprehensive search across 25+ clinical and genomics tables from National Library of Medicine (NLM) including conditions, drugs, genes, variants, ICD codes, and medical terminology | table, terms, count, offset, save_format |
| Clinical Trials Database Tools | clinical-trials-toolkit | — | Comprehensive clinical trials search using ClinicalTrials.gov API v2 with 400,000+ registered trials worldwide | tool_type, condition, intervention, phase, status |
| Medical Devices & Procedures Tools | medical-devices-toolkit | — | Medical device information from FDA (510k, PMA, recalls) and medical procedures/devices from SNOMED CT terminology | tool_type, query, limit, product_code, applicant |
Citations1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| CrossRef Search | crossref-search | — | DOI resolution, citation tracking, and comprehensive bibliographic data retrieval using CrossRef API | query, doi, author, title, journal |
Data Transformation1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Data Operations - Declarative Table Transforms | data-operations | — | Transform tabular workspace data (CSV/TSV/Excel) with a sequence of DECLARATIVE operations — no code. Compose: filter (subset rows), select (pick columns), derive_column (regex-extract a value or compute arithmetic), classify_regex (bucket a column into categories), group_by + aggregate (count/sum/mean/min/max/median/nunique per group), aggregate (whole-table), sort, dedup, limit, and join (against inline rows). Use this to turn raw tool/database output into the exact numbers a table or chart needs — e.g. count records per category, deduplicate by a key, extract a field from a coded string, or rank groups. Returns the resulting table plus row counts. | fileKey, operations, sheet |
Statistics1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Data Statistics - Statistical Analysis of Data Files | data-statistics | — | Run comprehensive statistical analysis on workspace data files. Computes descriptive statistics (mean, SD, median, CI), performs group comparisons (t-test, ANOVA, pairwise Tukey), correlation analysis, and data quality metrics (outliers, missing values). Supports multiple testing correction (Bonferroni, Benjamini-Hochberg). Use when the user needs statistical tests, group comparisons, or a numerical summary of their data. | fileKey, valueColumns, groupColumn, tests, multipleTestingCorrection |
Deseq21 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| DESeq2 - Differential Expression | deseq2-toolkit | — | Differential gene expression analysis from a count matrix using DESeq2 (Bioconductor), the standard method for RNA-seq. Fits a negative binomial model with shrinkage, giving log2 fold changes, p-values and BH-adjusted q-values per gene. Takes the count matrix from featurecounts-toolkit plus a list of sample conditions, and returns a ranked results table, a filtered significant-genes table, and normalised counts for plotting. THE step after quantification: counts tell you how much, DESeq2 tells you what changed. Runs on AWS ECS Fargate. | command, inputFile, conditions, outputFile, padjThreshold |
Ebi1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| EMBL-EBI Job Dispatcher Tools | ebi-job-dispatcher | — | BLAST (blastp, blastn, blastx, tblastn, psiblast) and other EMBL-EBI analyses, typically ~25 seconds. Returns percent identity, E-value and coverage. Set tool_name to 'ncbiblast' for BLAST. USE THIS FOR BLAST WHEN: the query is nucleotide/DNA/RNA, or the user needs a database outside Swiss-Prot — full UniProtKB/TrEMBL ('uniprotkb'), PDB structures ('pdb'), or nucleotide ('em_all'). For a PROTEIN query against Swiss-Prot — the common 'BLAST this protein' / 'what protein is this?' case — prefer smarts-blast instead: it is our own blastp and answers in about one second. Also runs multiple sequence alignment (clustalo, muscle, mafft), protein domains (iprscan5) and profile search (hmmer3_hmmscan), for which this is the right tool regardless. | action, job_id, tool_name, sequence, email |
Cross Database Search1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| EBI Search Cross-Database Discovery Tools | ebi-search | — | Comprehensive plugin for EMBL-EBI Search providing unified cross-database search across 170+ biological datasets with faceted filtering and cross-reference exploration | query_type, query, domain, entry_id, facet_fields |
Featurecounts1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| featureCounts - Read Quantification | featurecounts-toolkit | — | Count reads per gene from an aligned BAM/SAM against a GTF/GFF annotation, producing the count matrix that differential expression requires. This is the quantification step between alignment and DESeq2: HISAT2/STAR give you alignments, featureCounts turns them into per-gene counts. Auto-detects paired-end data (counting fragments, not reads) and reports an assignment summary so a zero-count run is visible rather than silent. Outputs both the native featureCounts table and a tidy two-column matrix ready for deseq2-toolkit. Runs on AWS ECS Fargate. | command, inputFile, annotationGtf, outputFile, featureType |
Bedtools1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| BEDTools - Genome Interval Analysis | bedtools-toolkit | — | Powerful suite for genome interval manipulation and analysis. BEDTools provides comprehensive functionality for intersecting, merging, counting, complementing and many other operations on genomic intervals in BED, GFF/GTF, VCF and BAM file formats. Essential for comparative genomics and functional annotation analysis. | command, inputFile, secondFile, genomeFile, outputFile |
Format Conversion1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Format Conversion Toolkit | format-conversion-toolkit | — | Comprehensive bioinformatics file format conversion toolkit. Converts between sequence formats (FASTA, FASTQ, GenBank, EMBL, AB1), alignment formats (SAM, BAM, CRAM, BED), variant formats (VCF, BCF), and annotation formats (GFF3, GTF, BED). Supports single files, multiple files, and automatic extraction of zip/tar archives. | conversionType, sourceFormat, targetFormat, inputFile, outputPrefix |
Genomicranges1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| GenomicRanges - Bioconductor Genomic Intervals | genomicranges-toolkit | — | Comprehensive genomic interval analysis using the Bioconductor GenomicRanges R package. Perform interval operations, overlap analysis, windowed analysis, and genomic annotation. | genomicRangesCommand, inputFile, outputFile, genomeBuild, chrPrefix |
Picard1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Picard Tools - High-Throughput Sequencing Processing | picard-toolkit | — | Comprehensive suite of Java-based command-line utilities for manipulating high-throughput sequencing (HTS) data and formats such as SAM/BAM/CRAM and VCF. Picard provides tools for data processing, quality control, metrics collection, and format conversion. | command, inputFile, outputFile, metricsFile, referenceSequence |
Sequence Editing1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Sequence Editor Toolkit | sequence-editor-toolkit | — | Edit, trim, translate, and manipulate DNA/RNA/protein sequences. Supports trimming, extracting regions, reverse complement, translation, transcription, quality filtering (FASTQ), masking, find/replace, and more. Intelligently chooses instant execution for small files (<5MB) or background processing for large files. | inputFile, operation, outputFile, startPosition, endPosition |
Vcftools1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| VCFtools Toolkit | vcftools-toolkit | — | Comprehensive VCF/BCF file manipulation and analysis toolkit. Provides filtering, statistics, format conversion, merging, annotation, and quality control capabilities for variant call format files. | command, inputFile, outputFile, outputFormat, referenceFile |
File Reading1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| File Reader - Read and Parse Data Files | file-reader | — | Read and parse bioinformatics data files from workspace. Supports Word (.docx), Excel (.xlsx), CSV, PDF, FASTA (.fasta, .fa, .fna, .faa, .ffn, .frn, .fas, .fsa), and GenBank formats. Includes automatic OCR for scanned PDFs via AWS Textract. Provides multi-stage data intake: metadata only, structure preview, or full content. Use for reading experimental data, lab reports, protocols, understanding file structure, or extracting specific information. | fileKey, mode, maxRows, sheet, extractTables |
File Search1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| File Search - Discover Files in Workspace | file-search | — | Use this tool when the user asks to find, search, list, show, or discover files in their workspace (e.g., 'find files with...', 'show me files that contain...', 'what files have...'). Searches by name, content keywords, metadata, format, topics, sections, and tags. Returns file metadata without reading full content. Essential for file discovery before using file-reader. | query, tags, format, keywords, topics |
File Writer1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| File Writer - Save Reports and Documents to Workspace | file-writer | — | Save text content, reports, and documents as files in the user's workspace. Use after generating a report to save it as a markdown file that users can view and download. Supports plain text and markdown format. | filename, content, folder |
Gatk1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| GATK Toolkit | gatk-toolkit | — | Genome Analysis Toolkit (GATK) for variant discovery and genotyping in high-throughput sequencing data. Executes GATK tools on AWS infrastructure with automatic file management and process monitoring. | gatkTool, inputBam, referenceFasta, outputPrefix, intervals |
Gnomad1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| gnomAD - Population Allele Frequencies | gnomad | — | Population allele frequencies for a variant from gnomAD (Genome Aggregation Database, Broad Institute) — ~800k exomes and genomes. Returns overall allele frequency, allele count/number, homozygote count, per-ancestry breakdown (African/African-American, Admixed American, Ashkenazi Jewish, East Asian, European Finnish/non-Finnish, Middle Eastern, South Asian), QC filters, and critically the FILTERING ALLELE FREQUENCY (faf95) used by ACMG/AMP criterion BA1/BS1. Accepts a variant id (17-43045677-T-C) or an rsID (rs429358). REQUIRED for deciding whether a variant is too common to be pathogenic — ClinVar and ClinGen say what a variant does, gnomAD says how often it occurs. | variant, dataset |
Gtex1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| GTEx - Tissue Expression | gtex-expression | — | Quantitative tissue expression (median TPM) for a human gene across ~54 tissues from the GTEx project. Resolves a gene symbol to a GTEx gencodeId and returns tissues ranked by median TPM, the top tissue, and a fold-change vs the next. Use for the 'Expression Profile' section of a gene analysis — the where/how-much dimension. | gene, gencodeId, datasetId |
Clinical Genomics1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| GWAS Catalog Disease Association Tools | gwas-catalog | — | Comprehensive plugin for NHGRI-EBI GWAS Catalog providing access to genome-wide association studies, SNP-trait associations, and disease genomics data | query_type, identifier, trait, variant_id, gene_name |
Interpro1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| InterPro Database Tools | interpro-toolkit | — | Comprehensive plugin for InterPro protein functional analysis including domain prediction, family classification, and functional site identification | tool_type, query, accession, protein_id, sequence |
Homer1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| HOMER - Hypergeometric Optimization of Motif EnRichment | homer-toolkit | — | Comprehensive suite for motif discovery, ChIP-Seq analysis, and peak annotation. Identifies enriched sequence motifs and analyzes transcription factor binding sites. | homerCommand, inputFile, outputDir, genome, motifLength |
Ncbi6 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| NCBI-Assembly | ncbi-assembly | — | Search genome assembly information from NCBI Assembly database | query, retmax |
| NCBI-Nucleotide | ncbi-nucleotide | — | Search and retrieve nucleotide sequences (DNA/RNA) from NCBI Nucleotide database | query, ids, organism, sequence_type, molecular_type |
| NCBI-Protein | ncbi-protein | — | Search and retrieve protein sequences from NCBI Protein database | query, ids, organism, protein_class, molecular_weight |
| NCBI-SRA | ncbi-sra | — | Search Sequence Read Archive from NCBI SRA database | query, retmax |
| NCBI-Structure | ncbi-structure | — | Search protein structures from NCBI Structure database with PDB integration | query, ids, organism, structure_type, experimental_method |
| NCBI-Taxonomy | ncbi-taxonomy | — | Search taxonomic information and organism classification from NCBI Taxonomy database | query, ids, rank, division, genetic_code |
Blast1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| NCBI-BLAST | ncbi-blast | — | BLAST against NCBI's own databases (nr, nt, swissprot, pdb). LAST RESORT — do not use this for an ordinary BLAST request. It submits to NCBI's public queue, which rate-limits this platform severely: a measured 30-residue peptide sat WAITING for over 15 hours and never completed, and results return as a background process to poll rather than an answer in the conversation. PREFER: smarts-blast for a protein query against Swiss-Prot (~1 second, ours), or ebi-job-dispatcher with tool_name 'ncbiblast' for nucleotide queries and other databases (~25 seconds). Use this tool ONLY when an NCBI-specific database such as nr or nt is explicitly required, and warn the user it may take hours. | sequence, program, database, expect, hitlist_size |
Clinvar1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| NCBI ClinVar - Curated Clinical Variants | ncbi-clinvar | — | Retrieve CURATED clinical variants for a gene from NCBI ClinVar (pathogenic / likely-pathogenic by default), with clinical significance, MOLECULAR CONSEQUENCE (frameshift/missense/nonsense/splice/…), protein change, HGVS, genomic position, and condition. Also returns a precomputed breakdown: consequenceCounts (e.g. {frameshift: 146, missense: 117, nonsense: 43, …}) and significanceCounts. Use THIS (not ncbi-dbsnp) for the clinical-variant section of a gene analysis — dbSNP returns raw, unannotated rs IDs. | gene, max, includeUncertain |
Dbsnp1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| NCBI-dbSNP | ncbi-dbsnp | — | Search SNP variant information from NCBI dbSNP database | query, retmax, saveToWorkspace, outputFormat, outputFilename |
Literature1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| NCBI-PubMed | ncbi-pubmed | — | Search PubMed literature database for scientific papers, articles, and reviews | query, retmax, sort, publication_type, date_range |
Open Targets1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Open Targets - Target-Disease Associations | open-targets | — | Ranked target-disease associations from the Open Targets Platform (EMBL-EBI / GSK / Sanger) — the canonical resource for drug target prioritisation. For a gene, returns the diseases it is associated with, each with an overall association score (0-1) and a breakdown by EVIDENCE TYPE: genetic association, somatic mutation, known drug, affected pathway, literature, RNA expression and animal model. Disease identifiers are MONDO. Use this to answer 'what diseases is this gene implicated in, how strongly, and on what evidence' — ClinVar gives individual variant pathogenicity, Open Targets gives aggregated gene-level evidence weighted across all sources. | target, size |
Structural1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| PDB Experimental Structure Database | pdb-toolkit | — | Access experimentally determined protein structures from X-ray crystallography, NMR, and cryo-EM. Best for high-resolution experimental structures and validation data. Use AlphaFold for predicted structures. | operation, pdb_id, query, search_type, format |
Msa1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| MSA - Multiple Sequence Alignment & Phylogeny | msa-toolkit | — | Multiple sequence alignment with MAFFT and phylogenetic tree inference with FastTree. Aligns a set of related DNA, RNA or protein sequences to reveal conserved and variable positions, and optionally builds an approximate maximum-likelihood tree from that alignment. Use for 'align these orthologs', 'which residues are conserved across species', 'build a phylogenetic tree', or comparing variants of a protein family. Auto-detects nucleotide vs protein. Distinct from BLAST, which finds similar sequences in a database one pair at a time — this aligns sequences you already have, all together. Tree output is Newick, which the platform's TreeViewer renders directly. Runs on AWS ECS Fargate. | command, inputFile, outputFile, sequenceType, alignmentMode |
Pipeline3 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| List Pipelines - Discover Predefined Bioinformatics Pipelines | list-pipelines | — | List all available predefined bioinformatics pipelines. Returns pipeline names, descriptions, required inputs, and estimated duration. Use this BEFORE building a pipeline from scratch to check if a predefined pipeline already covers the user's need. Can filter by category (variant-calling, transcriptomics, genomics, epigenomics, quality-control, alignment, structural-biology). | category |
| Pipeline Status - Check Multi-Step Pipeline Progress | pipeline-status | — | Check the status and progress of a running multi-step bioinformatics pipeline. Returns overall pipeline status, individual step statuses, progress percentage, and output files when complete. Use after start-pipeline to monitor pipeline execution. | executionId |
| Start Pipeline - Launch Multi-Step Bioinformatics Pipeline | start-pipeline | — | Start a multi-step bioinformatics pipeline. Supports two modes: (1) pipelineId mode — pass a predefined pipeline id from list-pipelines and provide only the input parameters; the steps are loaded automatically from the registry. (2) Custom steps mode — provide an explicit steps[] array for one-off pipelines. Returns an executionId to track progress with pipeline-status tool. | pipelineId, name, description, steps, parameters |
Process Management2 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Process Results - Get Process Output Files | process-results | — | Retrieve output files and results from a completed bioinformatics process. Use when a process status shows 'completed' and you need to access the generated files (e.g., PDB structures, VCF variants, analysis reports). Returns file paths, URLs, or file content for downstream analysis. | processId, includeContent |
| Process Status - Check Running Process Status | process-status | — | Check the status of a long-running bioinformatics process by process ID. Use when the user asks 'is my process done?', 'what's the status of process X?', or wants to check if a process has completed. Returns process status (running, completed, failed, cancelled), progress, and error messages if applicable. | processId |
Fpocket1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| fpocket - Binding Pocket Detection | fpocket-toolkit | — | Detect and rank candidate ligand-binding pockets on a protein structure using fpocket (MIT), a Voronoi tessellation / alpha-sphere method. Returns each pocket's location, volume, and a DRUGGABILITY SCORE (0-1) estimating how likely it is to bind a drug-like small molecule. Use for 'where could a drug bind this protein', identifying allosteric sites, or choosing a docking box. Complements molecular-docking: DiffDock-L docks blind without needing a pocket, fpocket tells you which cavities exist and which are worth targeting. Works on experimental PDB structures and AlphaFold models. Runs on AWS ECS Fargate. | command, inputFile, minSphereRadius, maxSphereRadius, minAlphaSpheres |
Pubchem1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| PubChem - Chemical Compound Lookup | pubchem | — | Look up a small molecule in PubChem (NIH), the largest public chemical database (119M+ compounds). THE ENTRY POINT for small-molecule work: resolves a common drug or chemical NAME to a structure (SMILES, InChIKey, formula) plus physicochemical properties (MW, XLogP, TPSA, H-bond donors/acceptors, rotatable bonds). Accepts a name ('aspirin'), PubChem CID, SMILES or InChIKey. Use this FIRST when the user names a compound but you need a structure to pass to docking, ADMET prediction or similarity search — ChEMBL is bioactivity-focused and is not a good name-to-structure resolver. | query, namespace, includeSynonyms |
Fastqc1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| FastQC - A quality control tool for high throughput sequence data | fastqc-toolkit | — | Quality control tool for high throughput sequencing data providing comprehensive quality assessment reports with summary graphs and statistics for raw sequence data. | command, inputFiles, outputDirectory, quiet, nogroup |
Fmlrc1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| FMLRC - FM-index Long Read Error Correction | fmlrc-toolkit | — | Long-read error correction tool using FM-index based methods to correct errors in noisy long reads using high-quality short reads as reference. | fmlrcCommand, inputFile, outputFile, shortReadsFile, kmerSize |
Trimmomatic1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Trimmomatic Toolkit | trimmomatic-toolkit | — | Comprehensive read trimming and adapter removal toolkit using Trimmomatic v0.39. Performs quality filtering, adapter trimming, and read preprocessing for both single-end and paired-end sequencing data. | command, inputFile, outputFile, outputFileR2, outputFileUnpairedR1 |
Visualization2 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Render File as Image — View as Picture | render-file-view | — | Render an EXISTING workspace file as a static image (PNG/SVG) that looks exactly like the interactive viewer — a heatmap/chart from a CSV, a slice of a FASTA sequence, a PDB structure, BAM coverage, or VCF variants. Use this to SHOW the user what their data looks like inline in chat. Pass just the file NAME (e.g. 'info_test.csv') — the tool resolves it to the workspace file automatically, so NEVER ask the user for a full storage key or path. The image is saved to the workspace and returned as a URL the response embeds. This renders a file the user already has; to generate a NEW chart from raw values, use the visualization-generator instead. | fileKey, format, region, chartType |
| Visualization Generator - Create Charts and Plots | visualization-generator | — | Generate professional charts, plots, and diagrams for scientific reports. Supports standard charts (bar, line, scatter, pie), specialized bioinformatics plots (volcano, MA, Manhattan, heatmap), workflow diagrams, and 3D molecular structures. Visualizations are saved to workspace and can be included in reports. | type, data, title, xLabel, yLabel |
Samtools1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| SAMtools Toolkit | samtools-toolkit | — | Comprehensive SAM/BAM/CRAM file manipulation toolkit using SAMtools v1.13. Provides format conversion, indexing, statistics, editing, and viewing capabilities for sequencing alignment data. | samtoolsCommand, inputFile, outputFormat, outputFile, referenceFile |
Scanpy1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Scanpy - Single-Cell RNA-seq | scanpy-toolkit | — | Single-cell RNA-seq analysis with Scanpy/AnnData: quality control and filtering, doublet detection (Scrublet), normalisation, highly-variable gene selection, PCA, optional Harmony batch integration, neighbour graph, UMAP, Leiden clustering, marker genes per cluster, and CellTypist cell-type annotation. Takes a .h5ad file, a 10x MEX directory, or a count matrix; returns an annotated .h5ad plus markers.csv, celltypes.csv and UMAP/QC plots. Use command 'full' for the whole workflow in one run, or an individual step to resume or re-tune. Handles up to 300,000 cells and refuses larger inputs with guidance rather than running out of memory. Runs on AWS ECS Fargate. | command, inputFile, species, minGenes, minCells |
Protein1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| SmartMatch Protein Search | smartsmatch-protein | — | FIRST CHOICE for identifying a protein sequence ('what protein is this?'). Fast similarity search over UniProt using vector embeddings, running on our own infrastructure — no external queue and no rate limit, unlike public BLAST. Returns UniProt accessions with a cosine similarity score, and finds remote homologs that alignment can miss. Note: this is embedding similarity, NOT an alignment — it does not produce percent identity or E-values, so follow up with ebi-job-dispatcher (tool_name 'ncbiblast') when alignment statistics are needed. | sequence, limit, threshold, include_metadata |
Tabix1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Tabix Query | tabix-query | — | Fast indexed genomic file queries using Tabix/TBI indexes. Query specific regions from compressed VCF, BED, GFF files without full extraction. Runs locally with HTTP range request support for efficient S3 access. | command, inputFile, indexFile, format, chromosome |
Web1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Tavily Web Search | tavily-search | — | Search the web for scientific literature and bioinformatics information using Tavily API | query, maxResults, searchDepth |
Ucsc1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| UCSC Genome Browser | ucsc-genome-browser | — | Access genome annotation and visualization data from the UCSC Genome Browser database | operation, genome, chrom, start, end |
Patents1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| USPTO Patents Search | uspto-patents-search | — | Search US patent database via PatentsView API for intellectual property research and prior art analysis | query, country_code, status, max_results, filing_date_start |
Vep1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Ensembl VEP - Variant Consequence Annotation | vep-toolkit | — | Annotate the functional consequences of variants in a VCF using Ensembl VEP (Apache-2.0). For each variant it reports the affected gene and transcript, the consequence term (missense, frameshift, stop gained, splice site, synonymous, intronic...), HGVS coding and protein notation, canonical-transcript flags, and optionally SIFT/PolyPhen scores and regulatory features. THE step between calling variants and interpreting them: GATK tells you a variant exists, VEP tells you what it does to the protein. Runs fully offline against a cached assembly on AWS ECS Fargate. | command, inputFile, outputFile, outputFormat, assembly |
Viennarna1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| ViennaRNA - RNA Secondary Structure | viennarna-toolkit | — | Predict RNA secondary structure with the ViennaRNA package (RNAfold, RNAsubopt, RNAplfold). Computes the minimum free energy fold and its dot-bracket notation, base-pair probabilities from the partition function, suboptimal structures within an energy window, and local accessibility profiles. Use for 'what structure does this RNA fold into', hairpins and stem-loops, riboswitch and UTR analysis, mRNA stability, or siRNA/guide accessibility. Temperature-adjustable. Runs on AWS ECS Fargate. | command, inputFile, outputFile, temperature, maxBpSpan |
Compression1 databases
| Database | Tool ID | Scale | Description | Key Parameters |
|---|---|---|---|---|
| Zip Toolkit | zip-toolkit | — | Comprehensive file compression and decompression toolkit supporting ZIP, GZIP, BZIP2, XZ, 7-Zip, and TAR formats. Handles bioinformatics files that are commonly distributed in compressed formats. | command, inputFile, inputFiles, compressionType, compressionLevel |
Discover Databases at Runtime
from smartsbio import SmartsBio
client = SmartsBio(api_key="sk_live_...")
tools = client.tools.list()
db_categories = [
"Sequence Search", "Protein & Structure Databases",
"Genomics & Variant Databases", "Pathway & Ontology",
"Clinical & Drug Databases", "Literature", "Patents",
]
for category in db_categories:
in_cat = [t for t in tools if t.category == category]
if in_cat:
print(f"\n{category}:")
for t in in_cat:
print(f" {t.id:35s} {t.description[:50]}")Query Examples
# UniProt — protein annotation
uniprot = client.tools.run(
tool_id="uniprot_toolkit",
input={"query": "BRCA1_HUMAN", "format": "json"},
)
print(uniprot["function"], uniprot["subcellular_location"])
# Ensembl — gene coordinates
ensembl = client.tools.run(
tool_id="ensembl",
input={"gene_id": "ENSG00000012048", "species": "human", "features": ["variants", "regulation"]},
)
# ChEMBL — drug-target bioactivity
chembl = client.tools.run(
tool_id="chembl_toolkit",
input={"target": "EGFR", "activity_type": "IC50", "limit": 20},
)
for compound in chembl["compounds"][:5]:
print(f"{compound['molecule_chembl_id']} IC50={compound['standard_value']} nM")
# ClinicalTrials.gov — open trials
trials = client.tools.run(
tool_id="clinical_trials_toolkit",
input={"condition": "breast cancer", "intervention": "BRCA1", "status": "RECRUITING"},
)
for trial in trials["studies"][:5]:
print(f"{trial['nct_id']}: {trial['brief_title']}")