Polymer informatics

Keywords Information systems Machine learning Ontology Polymer markup language Polymer informatics QSPR RDF Semantic web... 

The discipline of informatics can loosely be defined as the science of information and is invariably coupled with both the engineering of information systems and the practice of information processing. For our purposes, this defines the scope of what polymer informatics should be and therefore also the scope of this contribution. 

While considerable progress has been made in the area of small molecule informatics over the past several decades, any effort in the field of polymers has been timid at best and there is considerable scope for development. The main reason for the virtual non-existence of polymer informatics is the complex nature of polymers. This review will therefore start with an examination of the particular informatics challenges posed by polymers, in particular in the area of polymer representation and will also discuss some of the peculiarities of polymer information ( the science of information ). It will look at information systems for polymers ( engineering of information systems ) and a final section will review attempts to develop structure-property relationships for polymers ( practice of information processing ). The modeling of polymers either on the molecular - or meso-level - is outside the scope of this review. 

The first significant challenge that polymer informatics has to tackle, therefore, is to develop representations for both polymers and polymer data, which are computable, i.e., machine-comprehensible and to which rich metadata can be attached. 

Early efforts in what could be termed polymer informatics go back to an ACS symposium on polymer nomenclature in the late 1960s [46-51]. Papers in this symposium were mainly concerned with issues of polymer nomenclature and aspects of information retrieval. A first set of seminal papers only appeared about a decade later, as a consequence of another ACS symposium on the retrieval of polymer information in 1978 [52-59]. Collectively, the papers resulting from the 1978 symposium set out the challenges still faced by polymer informatics today the fuzzy nature of polymers and the variety of different types of descriptions and representations of polymers arising as a consequence, the problem of information... 

So far, all of the discussion in this review has focused on the representation of polymer structure and polymer information. However, another significant challenge in the development of polymer informatics is access to polymer data. In this context, the term access takes on two distinct meanings, namely access to data in terms of access-barriers (e.g., proprietary data, copyright considerations, etc.) and access in terms of the formats in which polymer data is communicated, handled and exchanged. 

Another aspect of polymer informatics, beyond the representation and registration of polymer information and data, is the conversion of data into knowledge and thus into the power to make decisions. To this end, the same tools, which are common in small molecule informatics, have also been used to study polymer data. The work that has been reported so far subdivides into two categories, namely classification and chemometrics problems and property prediction. 

Adams N, Murray-Rust P (2008) Engineering polymer informatics towards the computer-aided design of polymers. Macromol Rapid Commun 29 615-632... 

---