Using nanoinformatic methods to automatically identify optimum polymerosomes for drug delivery applications

D. GRILLO; E. MOCSKOS,; DAVID T. JONES; M. B. FERRARO; J. C. FACELLI

GALAPAGOS

Congreso; QUITEL 2014; 2014

Universidad de QUITO

In the last decades, there has been an increasing interest in the use of nanoparticles for drug delivery applications. Of special interest are polymeric devices, such as polymersomes and micelles [K. Letchford, H. Burt. Eur. J. Pharm. Biopharm. (2007), 65, 259?269]. The ability to encapsulate broad range of drugs and molecules, tunability and biocompatibility are some of the advantages that make these systems appealing as promising drug carriers. Many efforts to obtain nanoparticles with the desired characteristics have been done and a vast amount of empirical data of nanoparticle properties is now available, but very little is known about how to apply in silico approaches, commonly used in small molecule drug development, to improve the de novo design of these nanoparticles. The physical and chemical properties of these polymeric devices depend on several factors. In polymersomes, it has been shown that properties such as geometry, mechanical stability and permeability depend on the membrane properties, which are directly related with its chemical composition. Since these polymeric devices are intended to be used as drug delivery vectors, other important factors to be considered are the payload and the target, which are typically drugs and pathological cells, respectively. Due to the lack of well curated data bases on polymersomes physico chemical and bioactivity a prerequisite to develop quantitative predictive model of structure-function relationships is to efficiently obtain this data from the original literature. This task cannot be accomplished manually due to the large and complex literature volume. The work presented here show our efforts to develop methods to automatically extract information about chemical compositions (polymers), payloads (drugs) and targets (cells) by means of NLP (Natural Language Processing) techniques. For this task, a corpus of selected abstracts from PubMed has been prepared and each term associated with each category (polymer, drug, cell) is manually annotated. The annotated corpus is processed with the adequate NLP techniques to measure the performance and accuracy of this method. [PLOS One paper see: http://www.plosone.org/article/authors/info%3Adoi%2F10.1371%2Fjournal.pone.0083932;jsessionid=6A3D00B6CEEB2B3469C87283C3DD56FC