Pubmed Summariser

Workflow

When the user asks to summarise PubMed papers about a gene or disease:

1. Receive query: --query <term> or --demo (uses BRCA1)
2. esearch: Query https://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi for PMIDs
3. efetch: Fetch full XML records for those PMIDs
4. Parse XML: Extract title, authors, journal, date, abstract
5. Render output: Print terminal summary and write report.html

Algorithm / Methodology

• Query: <term> AND english[la], sorted by date descending, max 10 results (default)
• Author formatting: up to 3 authors as "Last FM", then "et al." if more exist
• Abstract: first sentence heuristic — split on . followed by uppercase letter, max 300 chars
• All NCBI requests include tool=clawbio&[email protected] per NCBI E-utilities policy
• Network timeout: 10 seconds

Output Structure

PubMed Research Briefing: <query>
================================
Found N papers (sorted by date, English only)

1. <title>
   Authors: <authors>
   Journal: <journal> | <date>
   Abstract: <first sentence>
   URL: https://pubmed.ncbi.nlm.nih.gov/<pmid>/

HTML report saved to <output>/report.html.

Dependencies

• requests (HTTP)
• xml.etree.ElementTree (stdlib — XML parsing)
• clawbio.common.html_report.HtmlReportBuilder (HTML rendering)

Safety

Every report includes the standard ClawBio medical disclaimer:

ClawBio is a research and educational tool. It is not a medical device and does not provide clinical diagnoses. Consult a healthcare professional before making any medical decisions.

Integration with Bio Orchestrator

Triggered by: "summarise PubMed papers about X", "recent papers on BRCA1", "research briefing", "gene papers", "disease papers"

Chaining partners: lit-synthesizer (broader literature), gwas-lookup (variant context), gwas-prs (polygenic risk)