Development of a General Algorithm to Identify n-Tuple Synthetic Lethal Reactions in Genome-Scale Models: A Systematic Approach Based on Fast-SL

Multi-drug resistance is known as a serious problem for curing the infections caused by microbial pathogens, which brings about high medical costs and mortality rates (Tanwar et al. ۲۰۱۴). To overcome this threat, it is suggested to take different subsystems of pathogens into consideration to be simultaneously attacked, instead of targeting only single essential reactions/genes (Silver ۲۰۰۷). Therefore, identification of new drug targets composed of elements related to different subsystems of pathogen(s) of interest is necessary. Thanks to GEnome-scale Metabolic models (GEMs) developed for these pathogens, it is possible to study the effects of reaction flux variations on their growth. As a result, these models make it possible to identify target sets (synthetic lethals), which their simultaneous removal prohibits the growth. Since, the number of potential cases, which should be examined to identify operative Synthetic Lethals (SLs), are surprisingly high (~۱۰۱۱ cases for quadruple SLs in exhaustive search), it is important to provide potent approaches capable of shortening the search space and the number of examined cases. Fast-SL (Pratapa et al. ۲۰۱۵) is the most recent and successful method presented to identify different SLs. This technique reduces the number of studied cases by approximately ۴۰۰۰-fold for finding triple SLs, compared to the exhaustive search. However, up to now, no comprehensive algorithms have been reported for this method and consequently computation of higher SLs (n > ۴) is not practical using the reported and available approach for Fast-SL, due to the necessity of developing new algorithms for each state. This issue has restricted further designed program packages due to the limitation in cardinality of SLs. In this work, a new comprehensive and integrated algorithm is provided based on the main idea of Fast-SL. This novel algorithm is capable to compute higher SLs without limitation on cardinality of the sets. Furthermore, by making minor revisions on original Fast-SL algorithm, the number of examined cases is reduced ۲-fold compared to the original method.

Keywords:

Systems biology , Drug target identification , constraint-based modeling