Python R
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


PdbMine: Mining the PDB database jointly with MEDLINE

Accompanying code and databases for Article Timely deposition of macromolecular structures is necessary for peer review, by Robbie P. Joosten, Hayssam Soueidan, Lodewyk Wessels and Anastassis Perrakis (Accepted for publication in Acta Crystallographica Section D).

This study starts with a list of PDB entries that you want to analyze. For the article, we took all PDB entries listed on the PDB website as of the 28 june 2013 at 15:25 Amsterdam time. Based on these entries, the corresponding MEDLINE entries are downloaded parsed and both the PDB and the MEDLINE entries are stored in a local SQLite3 database.

Getting the list of PDB entries

The PDB webservices are a bit troublesome to navigate in. The easiest way to get every entry with a corresponding 'citation' (e.g. a MEDLINE indexed article) is to use advanced seach.

Select "Citation" as a query type, then check filter to only keep "Primary citation", check the count, generate the results. Once results have been generated, export the table as a "Primary citation report".

Getting the PDB entries

PDB entries were obtained from PDB ftp site. For convenience, we provide a multi-thead FTP downloader. To use it, edit pdb-get-multi-thread.py, pointing to the correct list of PDB entries (first lines, CITATIONFILE variable) We also recommend to adjust the number of threads available depending on the machine used.

Once entries are downloaded, we clean the pdb-entries folder by removing any empty file:

find . -type f -empty | xargs rm

Inserting the entries in a local SQLite3 databases

Once the entries have been downloaded, we run medline-get.py then :

# Takes a while, as this will download all entries from MEDLINE corresponding to PDB entries 

Date processing

Once entries have been processed, parsed and inserted in the database, we compute the earliestpublication and earliestpublicdate by sourcing medline-get.py and then :


Results from these steps are stored in the provided SQLite database:local-medline-pdb.db. For convenience, the joined databaes are available as an R data.frame (actually as a data.table) in data/pdb.medline.RData.


All data aggregations and figure generation are performed in R and are described in deposit_time_analysis.Rmd

Impact factor table

For copyright reasons, this table cannot be made available. The impact factor table that we used correponds to the 2011 Thomson Reuters estimate.