A practical toolkit for finding, getting and evaluating cancer research, ordered roughly as you would use it.

Where to search

• PubMed is the core index. Most of the syntax tips below are for it.
• Europe PMC searches both abstracts and full texts in one place, so it catches details abstracts never mention, such as a cell line, a dose or an assay. Its Advanced Search can limit a query to one section of the paper, such as Methods, and SRC:PPR limits results to preprints.
• Google Scholar casts the widest net (theses, books, conference papers) and is the best tool for "Cited by" chasing.
• Citation-map tools (Connected Papers, ResearchRabbit, Litmaps, Semantic Scholar) take two or three key papers and show the cluster of work around them.
• bioRxiv and medRxiv have the newest work, which is not yet peer reviewed. SciELO (fully open access) and LILACS cover Latin American literature.

Cancer-specific resources

• Getting oriented: NCI's PDQ summaries (the health-professional versions are referenced and evidence-graded), and NCCN and ESMO guidelines for current standard of care.
• Data you can check yourself: cBioPortal (tumor genomics and survival), DepMap (CRISPR dependencies, drug sensitivity and metabolomics across hundreds of cell lines), COSMIC (somatic mutations), Human Protein Atlas (expression in tumors).
• Repurposing: the Anticancer Fund's ReDO_DB lists licensed non-cancer drugs with published evidence of anticancer activity. A companion open database, ReDO_Trials_DB, tracks repurposing clinical trials in oncology.
• Trials and regulators: ClinicalTrials.gov has results for many trials that were never published, and terminated trials count as evidence too. FDA review documents (Drugs@FDA) and EMA assessment reports (EPARs) contain more trial detail than the papers.
• Conference abstracts: AACR (Cancer Research supplements), ASCO (JCO supplements) and ESMO (Annals of Oncology supplements) have the newest data, but it is preliminary.

Search tips

• Build one synonym set per concept. Join terms within a set with OR, and join the sets with AND.
  • Disease: cancer, tumor/tumour, neoplasm, carcinoma, plus the specific type and its abbreviation (TNBC, NSCLC).
  • Target: the official gene symbol plus its aliases (HGNC lists them), and protein and isoform names.
  • Drug: generic name, code name and brand. Early preclinical papers often use only the code name (CB-839 before it was named telaglenastat). The NCI Drug Dictionary and PubChem list synonyms.
• Mix MeSH terms and free text. [mh] finds indexed papers well, and [tiab] catches the newest ones, because filters that depend on MeSH can miss records not yet indexed, like preprints and ahead-of-print papers.
• Wildcards: the asterisk needs at least four letters before it and switches off PubMed's automatic term mapping. For example, tumo*r* covers tumor, tumour and tumors.
• Proximity: "glutamine addiction"[tiab:~3] finds the two words in any order with up to three words between them. It only works in the title, title/abstract and affiliation fields, and not together with wildcards.
• Check what PubMed actually searched: on the Advanced page, expand the Search Details in your History.
• Vocabulary for different kinds of evidence:
  • Synergy: synerg*, "combination index", Bliss, Loewe, Chou-Talalay, isobologram
  • Repurposing: "drug repositioning"[mh], repurpos*
  • Animal studies: xenograft*, "patient-derived", orthotopic
  • Human data: clinical trial[pt], or systematic[sb] for systematic reviews
• Look for disconfirming evidence on purpose: terminated trials, phrases like "did not improve" or "no significant", and letters or comments on the paper.
• Set alerts so new papers come to you: PubMed's "Create alert" (it needs a free My NCBI account), plus Google Scholar alerts.

Example PubMed query that combines a target, a combination concept and a disease concept:

("glutaminase"[tiab] OR GLS1[tiab] OR telaglenastat[tiab] OR "CB-839"[tiab])
AND (synerg*[tiab] OR "combination index"[tiab] OR "drug combination*"[tiab])
AND (neoplasms[mh] OR cancer*[tiab] OR tumo*r*[tiab])

Getting full text legally

Roughly in order of effort:

1. Unpaywall browser extension (Chrome or Firefox): a green tab appears beside a paywalled article when a free, legal copy exists. Install it from unpaywall.org, because a Firefox add-on with the same name is a different tool that opens news sites through web archives.
2. Free full text filters: in PubMed, add free full text[Filter] or "pubmed pmc"[sb] to a search. Europe PMC's "Link to free full text" filter finds legal free copies hosted on other sites.
3. NIH-funded papers: anything accepted since July 1, 2025 has to be in PMC with no embargo from its publication date. Check PMC before giving up on a recent one.
4. Google Scholar's "All N versions" link often leads to a repository or author copy.
5. Preprints and author manuscripts: look on bioRxiv, the authors' university repository, or repository aggregators such as CORE and BASE.
6. Email the corresponding author with the title and DOI. Most reply within days, and sharing a copy for personal research is usually allowed. ResearchGate's "Request full-text" button works the same way.
7. Zotero's "Find Available PDF" fetches open-access copies automatically. Zotero also flags retracted papers in your library.
8. Theses (OATD, university repositories) often include the full methods and the negative results that didn't make it into the paper.

Reading cancer papers critically

• Cell lines: Cellosaurus flags misidentified and contaminated cell lines, including everything on the ICLAC register. One text-mining study of methods sections found about 8.6% of the cell lines mentioned were on the problematic list.
• Concentrations: compare the doses used in cell culture with plasma levels patients can actually reach. Metformin, for example, is often used at millimolar concentrations in culture, far above typical patient levels.
• Synergy claims: real synergy is measured across a matrix of doses and scored with a model such as Bliss, Loewe, ZIP or the combination index. A result like "A plus B worked better than A alone" does not show synergy.
• Culture media: standard media contain nutrient levels unlike those in the body. Metabolic dependencies, glutamine dependence in particular, can look different in animals or in physiological media such as HPLM or Plasmax.
• Replication: the Reproducibility Project: Cancer Biology completed 50 replication experiments from 23 high-impact papers, and the effects were on average 85% smaller than originally reported. Look for independent replication before building on a single result.
• Retractions and critiques: check PubPeer (its browser extension flags papers that are being discussed) and the Retraction Watch database, which is now openly available through Crossref.

Sources

• PubMed User Guide
• Section level search functionality in Europe PMC
• NIH Public Access Policy (Emory)
• BEIC on Springer Nature (CINCEL)
• Problematic cell lines and RRIDs (eLife)
• Reproducibility Project: Cancer Biology (COS)