Filtering of PDB Structure Download Statistics

The "raw" (original) and "filtered" (revised) statistics for web-based and FTP downloads of PDB data files (PDB, XML, and mmCIF) from all wwPDB sites and the logic used in filtering are available on this page. As of June 2009, wwPDB statistics are filtered to exclude instances of excessive downloads due to errors in external scripts.

Raw Statistics

The "Raw" statistics report all traffic to PDB files. This includes hits from web robots, manual downloads, internal network testing, and traffic from scripts accessing PDB files. In some months these numbers may be unusually high (for example, a runaway download script can cause excessive traffic from one IP address).

Filtered Statistics

The "Filtered" statistics report the number of downloads after all internal testing and known web robots traffic has been removed. To mitigate the impact of runaway scripts repeatedly downloading any given PDB ID in rapid succession, only up to 150 hits for each PDB ID from a single IP address per day are counted.

Additional filtering is performed on a case-by-case basis to remove less obvious statistical anomalies (for example, repeated downloads of the entire PDB archive within a few days in December 2007 by the same IP).

Summary Table of Filtered and Raw Download Statistics

This table reports the filtered and raw statistics on total file downloads from all wwPDB FTP sites (in the FTP column) and the sum of HTTP downloads and web page views from all wwPDB websites (in the HTTP column).

Known spikes in data downloads and views:

2007-08: first month after version 3.0/3.1 files were released

2009-03: first month after version 3.15 files were released