Axontologic :: Ai Connect

Axontologic's flagship product is Ai Connect, a web-based discovery platform for mining the biomedical scientific literature. Ai Connect uses the patented LaSSI and TIMI technologies to extract structure-activity relationships from internal and public research documents (millions of MEDLINE abstracts and over 800,000 compounds) and project them on to your chemical compounds. This allows the researcher to identify advantageous and deleterious bioactivities early in the research process.

Axontologic's approach to information analysis is such that our tools are clearly differen-tiated from modern search engine technologies. Search engines retrieve documents or web pages that must be then reviewed. We go a step further to extract the answers from the retrieved items by finding relevant sentences and pulling out the words or phrases that answer the question. For drug discovery applications, Ai Connect is preloaded with an understanding of proteins, genes, diseases and chemical structures. It finds latent statistical patterns between these concepts that could not be found with a search engine.

Ever wish that you could search PubMed, the patent literature and/or your own research documents with a chemical structure? Axontologic's TIMI technology makes searching text corpora with chemical structures easy. Simply import an existing structure or draw a new one and Ai Connect will retrieve the documents that are related to the components of your query structure.

For example, a query using the structure of the anti-inflammatory drug indomethacin will retrieve documents that mention other NSAIDS with structural similarity to indomethacin as well as similar compounds in other therapeutic categories. The published results of experiments involving similar compounds provide insight into the compounds your organization is developing. Studies at Merck Research Laboratories involving TIMI showed that indomethacin had structural similarity with another drug mentioned in the literature, clobazam, a 1,5-benzodiazepine used for treating anxiety. This similarity led to speculation that indomethacin might present CNS activity. In fact, indomethacin is known to aggravate depression, epilepsy and parkinsonism, though the mechanisms for those activities were unknown at the time of the studies. The results of these experiments were published in the Journal of Chemical Information and Computer Sciences (this paper is available on our Technology page).

Medicinal and computational chemists need to know which elements of their chemical compounds are responsible for activity and which might cause undesirable side effects or chemical properties. The ability to project those properties over their compounds allows them to visualize what can and should be modified to create a safe and effective drug.

In this example, a corpus of medicinal chemistry abstracts were analyzed with TIMI and used to identify 2D chemical descriptors correlated with efficacy in treating Crohns Disease. The atoms of the drug thalidomide are shown here colored by activity (red is highly correlated, blue is poorly correlated). These correlations were automatically extracted from the compounds described within the input abstracts.

Ai Connect can be used to explore relationships between the atoms and bonds of compounds and virtually any concept described in the corpus it analyzes.

Relevant publications describing our methods can be found on our Technology page.