Skills Hub.

Rank targets for a disease

Reads a disease's associated targets from Open Targets, splits each score by type of evidence and lists what supports the top one.

Input

Disease ID (EFO or MONDO)GWAS study (optional)

Output

Ranked targetsScores by evidence typeTop target's evidence

Open Targets

Map the clinical trial landscape

Pulls the ClinicalTrials.gov trials for a disease and summarizes them by intervention type, phase, sponsor and status.

Input

Disease name or termsIntervention, sponsor, phase or status (optional)

Output

Trial tableIntervention type by phaseSponsor summary

ClinicalTrials.gov

Check genetic constraint

Reads each gene's loss-of-function constraint from gnomAD, flags the genes that cannot tolerate it and rates each one's on-target risk.

Input

Gene symbols or IDs, typed or in a file

Output

Constraint scores per geneKnockout-tolerance tierFigures

gnomAD · MyGene.info

Tier a gene's clinical variants

Collects a gene's variants from ClinVar and CIViC, tiers each by actionability and maps them to therapies with clinical evidence.

Input

One human gene symbol

Output

Variants tiered by actionabilityLollipop plotTherapy matrix

ClinVar · CIViC · UniProt

Count alterations across cancer types

Pulls cBioPortal cohorts and shows how often your genes are mutated, amplified or deleted in each cancer type, with the mutation hotspots.

Input

Gene symbolsCohorts or cancer types (optional)

Output

Frequency by cancer typeHotspot and allele tablesFigures

cBioPortal

Assess druggability by modality

Scores a target's druggability for small molecules, antibodies and degraders from Open Targets and known drugs, and names the best-supported modality.

Input

Gene symbol or Ensembl IDDisease (optional)

Output

Scorecard per modalityKnown drugs and safety flagsFigures

Open Targets · UniProt

Decide whether to inhibit or activate

Reads each target's drug mechanisms and mouse phenotypes from Open Targets to call inhibit or activate, and flags where the evidence disagrees.

Input

Target genesIndication for each

Output

Call per target with confidenceEvidence behind each callConflict flags

Open Targets

Find where a target is expressed

Pulls a target's expression in each GTEx tissue, scores how tissue-specific it is and flags vital organs with high baseline expression.

Input

Gene symbol or Ensembl ID

Output

Expression per tissueTissue-specificity scoreOrgan safety flags

GTEx

Profile potency and selectivity

Gathers ChEMBL's curated IC50, Ki and Kd data for a compound across its targets, or ranks the compounds tested against one target.

Input

Compound name, ChEMBL ID or SMILESTarget ChEMBL ID, to rank its compounds

Output

Potency per targetSelectivity tableFigures

ChEMBL

Check drug-likeness

Standardizes your molecules, computes physicochemical properties and drug-likeness (Lipinski, Veber, QED), and flags structural alerts such as PAINS.

Input

SMILES list or fileNames (optional)

Output

Properties per moleculeDrug-likeness and alertsFigures

Predict off-target liabilities

Flags the safety-panel targets a compound may hit, by similarity to their known ligands, and checks each one against the compound's ChEMBL data.

Input

One compound (name, SMILES or ChEMBL ID)

Output

Safety-panel predictionsComparison with ChEMBL dataFigures

ChEMBL · UniProt

Train ADME models on your assays

Audits your assay table, trains and compares models for one ADME endpoint, and predicts new molecules with an uncertainty for each.

Input

Assay table with SMILES and one endpointDates, units or series (optional)

Output

Model comparisonHeld-out predictionsPredictions for new molecules

Model potency and find new scaffolds

Curates a target's ChEMBL potency data, benchmarks models on it and ranks your library's new scaffolds by predicted potency.

Input

Target symbol or ChEMBL IDCompound library (optional)

Output

Curated datasetModel benchmarkNew-scaffold candidates

ChEMBL

Map how a ligand binds

Finds the residues around a ligand in a PDB co-crystal structure, labels each contact by type and draws the pocket in 2D and 3D.

Input

PDB ID or structure fileLigand code (optional)

Output

Pocket contact tableInteraction diagram3D pocket view

RCSB PDB

Generate new molecules

Starts from your known actives, builds new candidates, keeps the novel ones and ranks them by predicted activity, drug-likeness and estimated ease of synthesis.

Input

TargetKnown actives and inactives (SMILES)

Output

Ranked new moleculesScores per objectiveFigures

Predict a protein's structure

Predicts a protein or complex structure from a sequence or UniProt ID and reports per-residue confidence, domain by domain.

Input

Sequence, UniProt ID or gene name

Output

Structure filesPer-residue confidenceConfidence by domain

UniProt

Design binders against a target

Prepares a target structure, designs protein binders or antibodies against a chosen site and ranks them by predicted interface confidence.

Input

Target structure (PDB ID or file)Hotspots or binder length (optional)

Output

Ranked binder candidatesSequences (FASTA)Figures

RCSB PDB

Humanize and assess an antibody

Numbers your antibody chains, flags sequence liabilities and likely immunogenicity, and proposes humanized versions scored for humanness.

Input

VH and VL sequences, or a FASTAReference antibody (optional)

Output

Constructs with liabilitiesImmunogenicity and humannessFigures

IEDB

Screen a library by docking

Prepares a receptor and your compound library, docks every compound, ranks the hits and groups them by scaffold.

Input

PDB ID or structure fileCompound library (SMILES or ChEMBL)

Output

Docking scores and top hitsScaffold clustersFigures

RCSB PDB · ChEMBL

Review the literature on a question

Searches peer-reviewed papers for your question, reads across them and writes a cited summary with an evidence table.

Input

A question or topicScope or time window (optional)

Output

Cited summaryEvidence table, one row per paper

Europe PMC · Unpaywall

Review the preclinical evidence

Searches the literature for a target and disease, pulls out the in vitro and in vivo findings and writes a cited summary of the gaps.

Input

Target and diseaseQuestions or time window (optional)

Output

Cited summaryEvidence table, one row per paper

Europe PMC · Unpaywall

Lay out a microplate experiment

Assigns treatments, replicates and controls to wells in a balanced, randomized layout that limits edge effects.

Input

Plate format, treatments and replicatesControls and covariates (optional)

Output

Plate mapsLayout tablesQuality report