PubChem Database Uses
What is the PubChem database used for?
PubChem is the U.S. National Library of Medicine's free, open chemistry database. People use it to look up information on chemicals: identity and structure, chemical and physical properties, biological activity, safety and toxicity data, patents and literature citations. Anyone can submit data to it and anyone can access it.
Approach: This comes from NLM's own pages: the PubChem training course, an NLM Technical Bulletin item, and a PubChem paper on NIH's PubMed Central. It describes the resource only. It does not cover any regulatory status PubChem may show for a particular substance.
What it holds and where the data comes from
- NLM, part of NIH, runs it. More than 1,000 sources supply the data, including academic labs, government agencies, chemical suppliers, pharmaceutical companies, journal publishers and individual researchers (NLM PubChem Training Course, Overview).
- It has three linked databases (PubChem paper on PubMed Central):
- Substance (SID): sample records as each depositor submitted them.
- Compound (CID): one record per unique chemical structure, built from the Substance records.
- BioAssay (AID): the public store of results from tests of chemicals for biological activity.
What people use it for
- Finding a chemical by name, molecular formula, structure or other identifier. You can also search by gene symbol, protein, biological pathway, organism, cell line or patent number (NLM Training Course index).
- Pulling together a chemical's properties, biological activity, safety and toxicity information, patents and literature in one place (PubChem). NLM calls it a key chemical information resource for scientists, students and the general public (NLM Technical Bulletin, 2024).
- Research work: linking compounds to test results, finding tests against a particular protein target, and using built-in analysis tools to identify probe chemicals and biological targets of interest (PubChem paper on PubMed Central).
- Downloading data and compound summaries, and following links to related sources such as clinical trials and published literature (NLM PubChem Training Course, Overview).
For regulatory work, PubChem is a reference that pulls together what other sources say. It is not a legal register. A substance's classification, Candidate List (SVHC) status or restrictions should be taken from the regulator's own instrument, such as ECHA's CLP Annex VI or REACH Annex XVII. If you have a particular substance in mind, I can look up its status in those registers.
Where do I find toxicity data on a compound page?
On a PubChem Compound Summary page, open the Contents menu at the top and use it to jump straight to the section you want (NLM PubChem Training Course, Download from a Compound Summary). I couldn't open PubChem's own pages this time, so I can't quote the exact name the compound page gives its toxicity section. Much of the toxicity content on those pages comes from the Hazardous Substances Data Bank (HSDB), and I could confirm how that record is organised.
Approach: I used NLM's PubChem training course for how to move around the page. For the toxicity headings I used the HSDB records held in PubChem, with acetophenone (CID 7410) as the example.
Toxicity headings in the HSDB record for acetophenone
You can also open this record directly from its HSDB source page in PubChem (https://pubchem.ncbi.nlm.nih.gov/source/hsdb/969):
- Toxicity Summary and Evidence for Carcinogenicity [1]
- Human Health Effects: Human Toxicity Excerpts; Skin, Eye, and Respiratory Irritations; Body Burden [2]
- Emergency Medical Treatment: Antidote and Emergency Treatment [3]
- Animal Toxicity Studies:
- Pharmacology: Metabolism/Pharmacokinetics; Interactions [7]
Related hazard data elsewhere in the same record
- GHS Classification and NFPA Hazard Classification [8]
- Occupational Exposure Standards: for example, the ACGIH TLV 8-hour TWA of 10 ppm, and a Workplace Environmental Exposure Level (WEEL) 8-hour TWA of 10 ppm (last revised 2007) [9]
Each entry gives the reference it came from. For example, the acetophenone occupational limits cite the ACGIH TLV documentation and the TERA WEEL list [9]. Use those references to go back to the original study or limit.
As with my earlier answer, the GHS entries shown in PubChem are a summary of what other sources report, not a legal classification. If you name a compound, I can pull its HSDB toxicity entries and check them against its CLP Annex VI harmonised classification.
Does it show GHS hazard classifications?
Yes. The PubChem record has a GHS Classification section, filed with the safety and hazard data, that lists pictograms, the signal word, hazard statements and precautionary statement codes. I checked the HSDB record for acetophenone (CID 7410) as an example [10][8].
Approach: I read the GHS section of the acetophenone HSDB record held in PubChem. I also checked OECD eChemPortal's entry for PubChem's HSDB.
What the section shows (acetophenone example)
- Pictograms: GHS05 (Corrosive) and GHS07 (Irritant) [10]
- Signal word: Danger [10]
- Hazard statements, each with its hazard class in brackets [10]:
- H227: Combustible liquid (Flammable liquids)
- H302: Harmful if swallowed (Acute toxicity, oral)
- H318: Causes serious eye damage (Serious eye damage/eye irritation)
- Precautionary statement codes: P210, P264, P264+P265, P270, P280, P301+P317, P305+P354+P338, P317, P330, P370+P378, P403 and P501. Each code links to its full wording on PubChem's GHS page (https://pubchem.ncbi.nlm.nih.gov/ghs/) [8]
- Next to it is a separate NFPA Hazard Classification (NFPA 704 diamond 2-2-0: health 2, fire 2, instability 0). It is a different system from GHS [8].
The section text explains GHS as the UN system that many countries have adopted and points to the PubChem GHS page for the full statements and categories [10].
How much weight to give it
- In the record I read, the GHS entries cite a reference number but don't say which body made the classification. So this is not a legal classification for any particular jurisdiction.
- OECD eChemPortal lists "HSDB at PubChem" as a participant source that does not provide classification and labelling (GHS) data to the portal [11].
- For a binding classification, use the regulator's own list: in the EU, the CLP Annex VI harmonised entry. National GHS lists, such as Japan's NITE GHS classifications or Korea's ICIS GHS records, serve the same purpose in those countries.
If you give me a CAS number, I can put PubChem's GHS entries next to the CLP Annex VI entry and the Japanese or Korean government classifications.
Can I download the data?
Yes. NLM says PubChem data can be downloaded three ways: individual summary download, programmatic download, and bulk download (NLM PubChem Training Course, Download Overview).
Approach: I used the "Download Summaries and Data" modules of NLM's PubChem training course. PubChem's own downloads help page (pubchem.ncbi.nlm.nih.gov/docs/downloads) would not open for me this time, so I can't give details of the bulk and programmatic routes.
From a Compound Summary page (Download from a Compound Summary)
- The Download button above the Contents menu gives you:
- the data used to build the page
- the 2D structure
- the 3D conformer
- The 3D conformer is computed to mimic a bioactive conformation. It is not experimental data.
- Individual sections also have their own Download button. The course's example is Literature > NLM Curated PubMed Citations. So you can download only the part you need, such as the safety and hazard or toxicity sections we discussed.
From search results (Download from Search Results)
- Search for a compound, then click Download on the right-hand side of the results.
- You can choose:
- Summary (Search Results)
- Chemical Structure Records
- Chemical Structure Images
- You pick a format such as CSV or JSON, and the file downloads straight away. The course doesn't list every format or say whether there are size limits.
Bulk and programmatic download
- NLM names both options and sends readers to PubChem's downloads help guide for details (Download Overview).
- Because that guide wouldn't open, I can't confirm from it the file-server address, the file formats, the programming interfaces or any usage limits.
As with the GHS entries, downloaded data is a compiled copy of what other sources report, not a legal classification.