PepDiana

Help

Use PepDiana without a learning curve

This guide explains the main workflows for browsing, filtering, saving, comparing, exporting, citing, and calling the local API.

Open API Docs

Complete user guide

From data layers to reproducible use

01

Understanding PepDiana data layers

Integrated Peptides are provenance-preserved identities from 26 public database resources. They occupy 25 stored source namespaces because AVPdb and mAVPdb share the AVPdb namespace, while HIPdb remains separately identified. Primary-Literature Peptides are formal sequence and modification-aware identities curated directly from papers. Evidence without sequence retains paper-scoped activity observations without inventing a canonical peptide sequence.

  • Use All for discovery across layers.
  • Use each layer tab when entity type matters.
  • Treat cross-layer sequence links as references, not merged annotations.
  • Integrated public IDs use PEPDIA-INT-xxxxxx; legacy CTPDB_Pxxxxxx aliases remain accepted for compatibility.

04

Inspecting evidence without sequence

These are scientifically useful activity observations whose paper did not provide a safely standardizable sequence in accessible text, table, figure, or structure context.

  • Open the mention page to preserve paper-local identity.
  • Open the stable activity-record URL for exact values and evidence text.
  • Do not include these records in FASTA or treat P1/PEP1 labels across papers as one molecule.

05

Interpreting positive, negative, and threshold records

Positive observations report activity under the stated assay context. Negative or threshold records preserve inactivity, lower/upper bounds, safety outcomes, or values outside the tested range.

  • Read comparator, raw value, normalized value, unit, assay, and target together.
  • Retain negative records for boundary and safety analyses.
  • Do not convert > or < values into exact measurements.

06

Downloading datasets

Versioned bulk assets include Integrated peptides, Primary-Literature Peptides, primary activity records, evidence-without-sequence mentions, cross-layer links, normalized activities, and computed properties.

  • Use manifest.json for row counts and SHA-256 checksums.
  • Use FASTA only for sequence-defined entities.
  • Record the release_version in every downstream analysis.

07

Using the PepDiana API

The REST API mirrors the three public data layers and provides stable list, detail, analytics, property, evidence, and download contracts.

  • Start with the site API Docs for curl, Python, and R examples.
  • Paginate list endpoints with page and page_size.
  • Prefer versioned downloads for large repeated analyses.

08

Using PepDiana Skill with agents

The PepDiana Skill maps natural-language questions to official API commands and labels every result with data_layer so an agent cannot mistake an activity record for a sequence-defined peptide.

  • Install or expose skills/pepdiana-api to the agent.
  • Use the eight public commands shown below.
  • Keep stable record URLs in the final answer for reproducibility.

09

Citation, license, and release notes

Cite PepDiana and the accessed release, retain source-database or publication attribution, and consult Changelog for data/model updates. Third-party source licenses continue to apply to contributed records.

  • Record the access date and release_version.
  • Use record-specific citation snippets where available.
  • Review Changelog before reproducing an older analysis.

Page guide

Browse is the primary search surface. It accepts sequence and keyword searches, dynamic activity/source options, active filters, pagination, sorting, and full-set exports.

Detail pages expose citation snippets, ProForma, HELM, SMILES, PubChem search, indexed sequence display, activity rows, matched primary experimental evidence, source records, and raw annotations.

Evidence searches the primary-literature evidence layer. It includes paper-level experimental records, negative or threshold observations, and evidence-only entries whose sequences are not yet safe to merge into canonical peptide records.

Favorites and Compare are local browser workspaces for shortlists. They support copy, remove, CSV/JSON/FASTA export, and side-by-side review.

Analytics summarizes the evidence landscape and provides clickable matrices for both integrated database records and independently curated primary-literature evidence.

API usage

GET/api/stats

Database-wide counts, source coverage, activity distributions, length bins, matrices, and co-occurrence summaries.

GET/api/peptides?page=1&page_size=20&keyword=dalbavancin&activity_type=Antimicrobial&sort=quality_desc

Paginated peptide list with keyword, sequence, length, multi-activity, multi-source, quality, mass, annotation, and sort controls.

GET/api/peptides/export?format=fasta&source_db=dramp&activity_type=Antimicrobial

Full matching-set export in CSV, JSON, or FASTA using the same filters as Browse.

GET/api/primary-peptides?page=1&page_size=20&activity_type=Antimicrobial

Search formally released Primary-Literature Peptides without mixing them into Integrated records.

GET/api/primary-peptides/{primary_peptide_id}

Retrieve one formal primary-literature peptide, release metadata, representations, properties, and cross-layer links.

GET/api/evidence/records?evidence_status=evidence_without_sequence&keyword=LL-37

Search primary-literature activity records, including evidence whose peptide sequence is unavailable in the source article.

GET/api/evidence/mentions/{mention_id}

Inspect one paper-scoped peptide mention, source location, records, and resolved peptide links.

GET/api/evidence/records/{activity_record_id}

Retrieve one stable experimental record with evidence text, raw/normalized values, assay context, parent mention, and paper.

GET/api/evidence/stats

Primary-literature layer sizes, evidence status, activity, target, measure, publication, result, and multifunctionality summaries.

GET/api/peptides/{peptide_id}/properties

Reported properties, versioned computed descriptors, standard representations, methods, assumptions, and failure states.

GET/api/peptides/search/by-sequence?q=ACDE

Motif or sequence search for exact subsequence discovery.

Use PepDiana with AI agents / Skills

PepDiana provides a lightweight agent Skill for programmatic workflows. The Skill teaches AI assistants to call the REST API directly, map natural-language requests to stable filters, and return compact peptide summaries with reproducible links.

Local package
skills/pepdiana-api
GitHub release

The Skill will be published from Iamzhangzixin after the public database deployment is ready.

The public Skill exposes eight entity-specific commands for Integrated search, Primary-Literature Peptide search, evidence search, stable detail retrieval, and evidence statistics. Every result names its data layer and includes reproducible detail links.

search-integrated

python skills/pepdiana-api/scripts/pepdiana_query.py search-integrated --activity-type Antimicrobial --min-quality 80 --limit 5

search-primary-peptides

python skills/pepdiana-api/scripts/pepdiana_query.py search-primary-peptides --activity-type Anticancer --limit 5

search-evidence

python skills/pepdiana-api/scripts/pepdiana_query.py search-evidence --evidence-status evidence_without_sequence --keyword LL-37 --limit 5

get-integrated-peptide

python skills/pepdiana-api/scripts/pepdiana_query.py get-integrated-peptide CTPDB_P010134

get-primary-peptide

python skills/pepdiana-api/scripts/pepdiana_query.py get-primary-peptide PEPDIA-PLP-000001

get-mention

python skills/pepdiana-api/scripts/pepdiana_query.py get-mention "PMC10043419::PEP000002"

get-activity-record

python skills/pepdiana-api/scripts/pepdiana_query.py get-activity-record "PMC10001067::ACT000001"

evidence-stats

python skills/pepdiana-api/scripts/pepdiana_query.py evidence-stats

Exports

Use current-page exports for quick inspection and All CSV/JSON/ FASTA for complete filtered sets. Detail pages export individual peptide records.

Browse exportCompare export

Citation

Suggested format for reports and reproducible analyses:

PepDiana: Multifunctional Peptide Database.
Release PepDiana-V1-20260620. Accessed YYYY-MM-DD at [deployment URL].

Individual peptide detail pages include record-specific APA and BibTeX snippets. Cite the original publication and retain source-database attribution when a record includes either.

Release notesAPI and download guidance

Quality ratings and scores

Integrated quality score (0–100)

This is a PepDiana internal completeness and traceability score, not an external database rating or a claim that a peptide is clinically effective. It combines sequence quality (30 points), source organism/taxonomy (15), activity evidence (25), literature support (15), and cross-database support (15).

Gold, Silver, and Bronze tiers

Gold:
score ≥70 with both activity and literature records.
Silver:
score ≥40 with at least one activity record.
Bronze:
all other records, including records whose provenance is useful but less complete.

The tier also checks required evidence fields, so the same numeric score can fall into a different tier when activity or literature support is absent.

Primary-literature evidence quality

Primary identities use a separate evidence grade. High means the released identity has no unresolved manual-review flag; Mediummeans the record is retained with a review flag that users should inspect in its source context. This grade does not replace assay-specific interpretation or independent experimental replication.

Good practice

  • Keep exported filter URLs with files for reproducibility.
  • Prefer Gold/Silver subsets when evidence confidence matters.
  • Check source records and literature before using peptides as benchmark labels.
  • Treat evidence-only and sequence-missing records as literature evidence, not canonical peptide entities.
  • Keep negative or threshold results when modeling activity boundaries or safety.