FTP Service

The PMC FTP Service provides source files for the articles in the PMC Open Access Subset; resources to enable cross-referencing PMC articles with identifiers such as PubMed IDs, DOIs, Manuscript IDs, ISSN, etc.; and other resources. The contents of the PMC FTP site is normally updated once a day.

The base URL of the FTP site is ftp://ftp.ncbi.nlm.nih.gov/pub/pmc. If you have questions or comments about this service, please write to the PMC help desk.

Article files from the PMC Open Access Subset

This FTP service may be used to download the source files for any article in the PMC Open Access Subset. There are two file formats provided:

.tar.gz - these are archive files that include all of the source material for the article, including:

  • A .nxml file, which is XML for the full text of the article, encoded in the NLM Journal Archiving and Interchange DTD.

  • Image files from the article, and graphics for display versions of mathematical equations or chemical schemes.

  • Supplementary data, such as background research data or videos.

  • PDF, if available

  • Converted video files, in a number of formats, suitable for streaming on the web. These files have the suffix, "-pmcvs_normal" to distinguish them from original, publisher-supplied files.

.pdf - the PDF associated with the article (same as that in the .tar.gz file). Note that not every article has a PDF.

Because there are so many articles, in order to prevent any one FTP folder from having thousands of files, these .tgz and .pdf files are distributed randomly in a two-level-deep structure. For example, the files for the article PMC13901 are (randomly) assigned into the folder b0/ac/. For any particular article, there are two ways to discover the location of its files: using one of the file lists (described next) or using the OA web service.

Using the file lists

There are several files on the FTP site that provide lists of the content that is available: file_list.txt, file_list.csv, file_list.pdf.txt, and file_list.pdf.csv. These provide indeces (in one of two formats) of open-access articles available from PMC.

file_list.txt

This file can be retrieved from ftp://ftp.ncbi.nlm.nih.gov/pub/pmc/file_list.txt. The first line of the file gives the date and time at which it was last generated. Every subsequent line contains information about one article in PMC. For example:

b7/8b/Open_Ophthalmol_J_2009_Jun_11_3_26-28.tar.gz	Open Ophthalmol J. 2009 Jun 11; 3:26-28	PMC2701320

This line is divided into three main fields, delimited by tab characters. Those fields are:

  1. The fully qualified name of the .tar.gz file for an article.

  2. The article citation, comprising:

    • journal title abbreviation
    • publication date
    • volume
    • issue
  3. The PMC accession number

file_list.csv

This file can be retrieved from ftp://ftp.ncbi.nlm.nih.gov/pub/pmc/file_list.csv.

Contents: Fields are the same as above, except separated by commas, and with the addition of a trailing timestamp indicating the last update to the article in PMC. For example:

8d/2f/Int_J_Health_Geogr_2003_Sep_25_2_7.tar.gz,Int J Health Geogr. 2003 Sep 25;2:7,PMC222916,2011-05-06 15:30:22

file_list.pdf.txt

This file can be retrieved from ftp://ftp.ncbi.nlm.nih.gov/pub/pmc/file_list.pdf.txt.

This is the same as the file_list.txt file, but it only lists those articles that have PDFs, and, of course, it gives the location of the PDF rather than the location of the .tgz file.

file_list.pdf.csv

This file can be retrieved from ftp://ftp.ncbi.nlm.nih.gov/pub/pmc/file_list.pdf.csv.

This is the same as file_list.pdf.txt, except uses commas as delimiters, and adds the trailing timestamp, indicating the last update to the article in PMC.

To find an article from PMC on the FTP site

Copy the PMC accession number from the PubMed Central URL (for example, in the URL http://www.ncbi.nlm.nih.gov/pmc/articles/PMC13901/, the accession number is PMC13901), and then search for that accession number in the file_list.txt file.

Alternatively, you can use this accession number with the OA web service, by retrieving information from, for example, http://www.pubmedcentral.nih.gov/utils/oa/oa.fcgi?id=PMC13901.

Obtaining DOIs and PubMed IDs for articles in PMC

Use PMC-ids.csv.gz to associate PMC articles with a PMC accession number, a PubMed ID, and the corresponding DOI.

PMC-ids.csv.gz is a comma-delimited file with the following fields:

  • Journal Title
  • ISSN
  • Electronic ISSN
  • Publication Year
  • Volume
  • Issue
  • Page
  • DOI (if available)
  • PMC accession number
  • PubMed ID (if available)
  • Manuscript ID (if available)
  • Release Date (Mmm DD YYYY or live)

Format:

Journal Title,ISSN,eISSN,Year,Volume,Issue,Page,DOI,PMC accession number,PMID,Manuscript Id,Release Date

Sample entries:
  • Mol Biol Cell,1059-1524,1939-4586,2000,11,6,2019, ,PMC14900,10848626, ,live
  • J Neurosci,0270-6474,1529-2401,2005,25,24,5740,10.1523/JNEUROSCI.0913-05.2005,PMC1201448,15958740,NIHMS3372,live
  • Cancer Res,0008-5472,1538-7445,2007,67,17,8022,10.1158/0008-5472.CAN-06-3749,PMC1986634,17804713,NIHMS25090,Sep 1 2008
  • Proc Natl Acad Sci U S A,0027-8424,1091-6490,2007,104,43,17075,10.1073/pnas.0707060104,PMC2040460,17940018, ,live
  • Cell Host Microbe,1931-3128,1934-6069,2007,2,6,404,10.1016/j.chom.2007.09.014,PMC2184509,18078692,NIHMS36164,live
  • Proc Natl Acad Sci U S A,0027-8424,1091-6490,2008,105,21,7382,10.1073/pnas.0711174105,PMC2396716,18495922, ,Nov 27 2008
  • PLoS Med,1549-1277,1549-1676,2008, ,Immediate Access,e168,10.1371/journal.pmed.0050168,PMC2494565,18684010, ,live

Note:

  • If any information is not available, entries will contain an empty space.
  • Articles that show a Release Date are under embargo and not yet available on the PMC public site.
  • When embargoed articles are released to the PMC public site, the Release Date field value changes to "live".

XML for data mining via FTP

The files below contain XML (and only XML) for all the articles in the PMC open access subset. These files were created for users who need PMC XML for data mining and processing purposes, but do not need PDFs, images, or supplementary data.

Note that these files are quite large (each greater than one gigabyte).

Suggested FTP client configuration

After a series of experiments using ftp clients with NCBI's ftp server, we've found that the configuration of ftp clients can seriously affect performance.

Note:

NCBI recommends setting the TCP buffer size to 32Mb.

For more information on FTP configuration, please see the US Department of Energy's Guide to Bulk Data Transfer over a WAN.

Last updated: Mon, 23 Dec 2013