Haibin Liu

Alzheimer's disease (AD) is one of the most complicated progressive neurodegeneration diseases that involve many genes, proteins, and their complex interactions. No effective medicines or treatments are available yet to stop or reverse the progression of the disease due to its polygenic nature. To facilitate discovery of new AD drugs and better understand the AD neurosignaling pathways involved, we have constructed an Alzheimer's disease domain-specific chemogenomics knowledgebase, AlzPlatform (www.cbligand.org/AD/ ) with cloud computing and sourcing functions. AlzPlatform is implemented with powerful computational algorithms, including our established TargetHunter, HTDocking, and BBB Predictor for target identification and polypharmacology analysis for AD research. The platform has assembled various AD-related chemogenomics data records, including 928 genes and 320 proteins related to AD, 194 AD drugs approved or in clinical trials, and 405,188 chemicals associated with 1, 023,137 records of reported bioactivities from 38,284 corresponding bioassays and 10, 050 references. Furthermore, we have demonstrated the application of the AlzPlatform in three case studies for identification of multitargets and polypharmacology analysis of FDA-approved drugs and also for screening and prediction of new AD active small chemical molecules and potential novel AD drug targets by our established TargetHunter and/or HTDocking programs. The predictions were confirmed by reported bioactivity data and our in vitro experimental validation. Overall, AlzPlatform will enrich our knowledge for AD target identification, drug discovery, and polypharmacology analyses and, also, facilitate the chemogenomics data sharing and information exchange/communications in aid of new anti-AD drug discovery and development.

Hurricanes can cause extensive power outages, resulting in economic loss, business interruption, and secondary effects to other infrastructure systems. Currently, power companies are unable to accurately predict where outages will occur. Therefore, it is difficult for them to deploy repair personnel and materials, and make other emergency response decisions in advance of an event. This paper describes negative binomial regression models for the number of hurricane-related outages likely to occur in each one square kilometer grid cell and in each zip code in a region due to passage of a hurricane. The models are based on a large Geographic Information System database of outages in North and South Carolina from three hurricanes: Floyd (1999), Bonnie (1998), and Fran (1996). The most useful explanatory variables are the number of transformers in the area, the company affected, maximum gust wind speed, and a hurricane effect. Wind speeds were estimated using a calibrated hurricane wind speed model. Pseudo R-squared values and other diagnostic statistics are developed to facilitate model selection with generalized negative binomial models.

This paper introduces a new method for estimating the time at which electric power will be restored after a major storm. The method was applied for hurricanes and ice storms for three major electric power companies on the East Coast. Using an unusually large dataset that includes the companies' experiences in six hurricanes and eight ice storms, accelerated failure time models were fitted and used to predict the duration of each probable outage in a storm. By aggregating those estimated outage durations and accounting for variable outage start times, restoration curves were then estimated for each county in the companies' service areas. The method can be applied as a storm approaches, before damage assessments are available from the field, thus helping to better inform customers and the public of expected post-storm power restoration times. Results of model applications using testing data suggest they have promising predictive ability.