Drug discovery using very large numbers of patents. General strategy with extensive use of match and edit operations

Abstract

A patent data base of 6.7 million compounds generated by a very high performance computer (Blue Gene) requires new techniques for exploitation when extensive use of chemical similarity is involved. Such exploitation includes the taxonomic classification of chemical themes, and data mining to assess mutual information between themes and companies. Importantly, we also launch candidates that evolve by "natural selection" as failure of partial match against the patent data base and their ability to bind to the protein target appropriately, by simulation on Blue Gene. An unusual feature of our method is that algorithms and workflows rely on dynamic interaction between match-and-edit instructions, which in practice are regular expressions. Similarity testing by these uses SMILES strings and, less frequently, graph or connectivity representations. Examining how this performs in high throughput, we note that chemical similarity and novelty are human concepts that largely have meaning by utility in specific contexts. For some purposes, mutual information involving chemical themes might be a better concept.

The Concept of Novel Compositions of Matter: A Theoretical Analysis

“…In a 2011 study [17], the author and colleagues applied somewhat unusual molecule mining techniques and drew the conclusion that "chemical similarity and novelty are human concepts that largely have meaning by utility in specific contexts. For some purposes, mutual information involving chemical themes might be a better concept".…”

“…When undertaking the previous study [17], applying rigorous substructure similarity tests to very large collections of SMILES strings, and especially for similarity between many members of the collection rather than a single query, can be a rate limiting step for larger molecules in a workflow. We originally sought to speed this by first using a highly optimized industry standard for pattern matching, the regular expression ("regex") [20] in the context of the Perl language [21].…”

“…"Methods" are included for brevity because the algorithms and hence software are essentially little more than basic informationtheoretic formulae that quantify the following considerations; further required theoretical and methodological details are given in [17], but enough explanation will be made available to understand the principles. The considerations are, in the first few Sections, some different but interrelated perspectives of how much information a claim for a novel composition of matter should contain, and what that means.…”

“…It may be considered a context dependent effect, meaning that the electronic effects of the surrounding structure can have significant chemical effect on a substructure capable of existence as a molecule in its own right. That is, allowing for addition and removal of hydrogen atoms: hydrogen atoms are not explicit in the study, they are there by implication to satisfy free valencies [17]. The problem is to some extent a point of view, but not totally.…”

“…A biochemist would consider adenosyl diphosphate and adenosyl triphosphate as very different, but then so would a quantum chemist. Nonetheless, the simple fact that it is possible to identify substructure or fragments as recurrent chemical themes and associate them differentially with companies [17] indicates that the notion of distinct chemical themes as modular components makes sense. Our previous study involved automatic generation of taxonomy of chemical themes, because a question of what is a theme is formally a matter of whether we pool distant, close, or immediate family members.…”