Table 1.

Comparison of state-of-the-art PI extraction programs (Kabiljo, Clegg, and Shepherd, 2009)

PI programs	iHOP	AkanePPI	OpenDMAP	Protein Corral	Bayesian Networks Method	PIMiner
Source	Hoffmann and Valencia, 2005	Saetre, Kenji, and Tsujii, 2008	Hunter et al., 2008	Rebholz-Schuhmann et al., 2008	Chowdhary, Zhang, and Liu, 2009	This Study
Webserver / webutility / standalone	Webserver and webservice	Standalone	Standalone	Webserver and webutility	Standalone	Webserver and webutility
Ease of installation, configuration and usage	Easy	Difficult requires installation of several, non-trivial components	Difficult configuration requires XML configuration file, not supplied	Easy webservice requires some basic coding	Difficult	Easy
Easily testable on different training/test data	Difficult does not accept user-submitted text	Difficult requires linguistic expertise	Difficult requires custom-written sample code from authors	Difficult does not allow pretagged text, considers only entities normalized to Uniprot	Difficult	Easy Module B designed specifically for this purpose
Allows user training data	No	Yes	Yes	No	No	Yes
Allows user raw text (test/query data) as input	No	Yes	Yes	Yes (in Whatizit version)	Yes	Yes
Allows user NER tagging	No	Yes	Yes	No	No	Yes
Easy integration with user programs	Yes	Yes	Yes	Yes	Yes	Yes
User specified interaction types	No	No	No	No	No	Yes