Peptide research trends

This page counts papers. It does not measure whether anything works. Publication volume is descriptive bibliometrics: it tells you how much scientific attention a compound has received, and nothing else. A peptide with 4,000 papers is not more effective, safer, or better evidenced than one with 40. Attention follows commercial funding, patentability, regulatory interest, and the accident of which lab picked a molecule up in 1994.

Counts also mix study types without distinction. A single rodent pilot and a 20,000-patient cardiovascular outcomes trial each contribute one record. For evidence quality rather than evidence quantity, read the graded evidence matrix on each compound profile and our methodology. Nothing here is medical advice or a recommendation to use any compound.

Peptide publishing has risen more or less continuously since 2000, roughly 2.6x over the period, and 2025 was the largest year on record with 37,372 records. The final bar is faded because it is structurally incomplete: PubMed indexing lags publication by months, so the current year always looks like a collapse and never is one. Read the trend up to 2025 and treat anything after it as provisional.

The head of this list is not a ranking of peptide therapeutics. It is dominated by molecules the human body already makes, which carry decades of basic-biology research that has nothing to do with administering them, and by approved drugs with large industry trial programs. Rows tagged ENDOGENOUS fall into the first group.

Drop those and you get the list most readers are actually after: compounds people take, ranked by how much has been published about them. It is worth noticing what fills the top of it. These are approved pharmaceuticals with decades of trial programs behind them, not the compounds that dominate peptide forums. The gap between those two lists is the point of this dataset.

This is the more useful half of the dataset. These compounds circulate widely in peptide communities and vendor catalogues, and the published literature behind them is close to nothing. A handful of papers cannot establish a safety profile, an effective dose, or a mechanism in humans. Where a count looks implausibly low, check the query in the download first: some of these compounds are studied under names our synonym list does not carry.

Publications in the last five complete years divided by the five before them. Compounds with fewer than 15 papers in the earlier window are excluded, because a jump from two papers to eight is a 4x ratio and means nothing.

Free to download, reuse, and republish under a Creative Commons BY 4.0 license. The JSON carries everything: per-year counts, the exact esearch query string for each compound, the synonyms we excluded and why, and the two diagnostic columns. The long CSV is one row per compound per year, which is the shape most plotting tools want.

Counts come from the NCBI PubMed E-utilities esearch endpoint, pulled on 2026-07-23 by scripts/build-pubmed-dataset.mjs in our public build pipeline. Raw responses are cached so the dataset can be rebuilt without re-querying NCBI.