SignaLink 3 (self-curated) (http://signalink.org/)
Fazekas et al. 2013, Korcsmáros et al. 2010

A database assigning proteins to signaling pathways using the full texts of pathway reviews. Compared to most signaling resources, SignaLink uses more than 20 review papers per pathway on average. It aims at reducing data and curation errors during the curation process.

SignaFish (http://signalink.org/)
Csályi, Fazekas, et al. 2016

A Zebrafish-Specific Signaling Pathway Resource. The first integrated resource to analyze signaling pathway cross-talks, miRNAs and regulatory enzymes in zebrafish.

ACSN (https://acsn.curie.fr)
Kuperstein et al. 2015, Calzone et al. 2008

The original sources of information come from review articles in high-impact journals. The information is extracted from these papers and represented in the form of biochemical interactions. This map is enriched with data from recent discoveries. It is necessary for the represented biochemical process to have evidences from more than two studies.

InnateDB (http://www.innatedb.com/)
Breuer et al. 2013, Lynn et al. 2008, Lynn et al. 2010

Manually curated binary protein interaction resource. A platform designed to represent systems-level mammalian innate immune response analyzes. It is a comprehensive database of human, mouse and bovine molecular pathways and interactions curated from public molecular interaction databases, and is not only limited to innate immunity-relevant data. Data is supplied with UniProt IDs, PubMed references and experimental evidences.

Reactome (http://reactome.org/)
Croft et al. 2014, Haw et al. 2011, Croft et al. 2014

It is not possible to extract binary information programmatically from its dataset. The curation method doesn’t include binary interactions, the available lists are based on automatic expansion of reactions and complexes, which makes the data unreliable. Also, it doesn’t make it possible to assign references to interactions.

Signor (http://signor.uniroma2.it/)
Perfetto et al. 2016

Signaling Network Open Resource aims to store and organize signaling information published in scientific literature in a structured format, as binary causative relationships between biological entities. Relationships are provided with literature reporting experimental evidence. The contained data is mapped to the human proteome.