2016 INFORMS Annual Meeting Program

TA16

INFORMS Nashville – 2016

2 - Some Thoughts On Implementing Linear Optimization Algorithms Tamás Terlaky, Lehigh University, terlaky@lehigh.edu We discuss some aspects of implementing simplex and interior point algorithms for LO. Among others, we discuss the role of duality in algorithms design, in preprocessing, in choosing the most promising algorithm, in choosing if the given problem should be considered as primal or dual. The provocative question naturally arises: Do we indeed have both primal and dual simplex algorithms? Another question is if an optimal basis is always needed? Then, the algorithmic and practical consequences of choosing either crossover or optimal basis identification procedure are discussed. 3 - An Advanced Starting Basis For The Simplex Algorithm Nikolaos Ploskas, Carnegie Mellon University, Pittsburgh, PA, United States, nploskas@andrew.cmu.edu, Nikolaos Samaras, Nikolaos Sahinidis The computation of a starting basis for the simplex algorithm is of great importance. We propose six algorithms for constructing an initial basis using various ordering methods in order to generate a nearly-triangular and sparse initial basis. We give the initial bases as input to the CPLEX solver and compare the performance of the primal and dual simplex algorithm using the proposed algorithms against CPLEX advanced starting basis and crash procedures. The best algorithm results in 95% and 87% reduction of the execution time of the primal and the dual simplex algorithm, respectively. 4 - A Planning Approach And a Decision Support System For Wine Bottling Operations Alfredo R. Squadritto, Pontificia Universidad Católica de Valparaíso, School of Industrial Engineering, Valparaíso, Chile, squadritto@ucv.cl, Ernesto Vásquez, Ricardo A. Gatica, Sergio G. Flores, Ricardo Gatica We will describe a MIP-heuristic based decision support system (DSS) for production planning and scheduling at wine bottling plants. The production system involves several complicating features such as the coexistence of items made-to-order with items made-to-inventory, a complex structure, several constraints on wine supply, and a variety of company’s policies. The DSS offers tools for easily modifying and analyzing the production schedule using a friendly graphic user interface. It’s been implemented in 3 plants. The largest plant handles about 2400 SKUs, 400 intermediate products, and 3500 types of raw material, and it processes, on average, 250 production orders per week on 2 parallel production lines. Centre/University of Toronto, Toronto Medical Discovery Tower 11-311, 101 College Street, Toronto, ON, M5G 1L7, Canada, michael.hoffman@utoronto.ca 1 - Modeling Methyl-sensitive Transcription Factor Motifs With An Expanded Epigenetic Alphabet Michael M Hoffman, Scientist, Princess Margaret Cancer Centre, Toronto Medical Discovery Tower 11-311, 101 College St, Toronto, ON, M5G 1L7, Canada, michael.hoffman@utoronto.ca Michael M Hoffman, Scientist, University of Toronto, Toronto Medical Discovery Tower 15-701, 101 College St, Toronto, ON, M5G 1L7, Canada, michael.hoffman@utoronto.ca To understand the effect of DNA methylation on gene regulation, we developed methods to discover motifs and identify TF binding sites (TFBS) in DNA with covalent modifications. Our models expand the standard A/C/G/T alphabet, adding m for 5-methylcytosine. We adapted the position weight matrix model of TFBS affinity to an expanded alphabet. Using ChIP-seq data from Mouse ENCODE and others, we identified modification-sensitive cis-regulatory modules. We elucidated various known methylation binding preferences, including the methylation preferences of ZFP57, C/EBP , and c-Myc. 2 - Data, Informatics, And Analytical Challenges In Genomic Medicine Elizabeth A. Worthey, Faculty Investigator & Director, HudsonAlpha Institute for Biotechnology, 601 Genome Way, Huntsville, AL, United States, lworthey@hudsonalpha.org Application of Next Generation Sequencing has transformed genomic research. It is also transforming medicine through use as a molecular diagnostic test in both rare disease and oncology. It can also identify much (though not all) of the variation associated with more common and polygenic disease. To support such clinical and translational advances investment in computational applications, hardware, and methodologies has been necessary. Informatics environments, TA14 104D-MCC Data Mining in Genetics and Genomics Sponsored: Data Mining Sponsored Session Chair: Michael M Hoffman, Princess Margaret Cancer

tools, and processes supporting medical genomics have been developed or refined and validated for clinical use. This talk will highlight various successes and will

discuss potential solutions to some of the challenges that remain. 3 - Machine Learning For Predicting Vaccine Immunity Eva K. Lee, Georgia Institute of Technology, eva.lee@isye.gatech.edu

This work is joint with Emory Vaccine Center and CDC. The ability to better predict how different individuals will respond to vaccination and to understand what best protects individuals from infection greatly facilitates developing next- generation vaccines. It facilitates both the rapid design and evaluation of new and emerging vaccines as well as identifies individuals unlikely to be protected by vaccine. We describe a general-purpose machine learning framework, DAMIP, for discovering gene signatures that can predict vaccine immunity and efficacy.

TA15 104E-MCC

Academic Job Search Panel Invited: INFORMS Career Center Invited Session Chair: Wedad Jasmine Elmaghraby, University of Maryland, College Park, MD, United States, welmaghr@rhsmith.umd.edu 1 - Academic Job Search Panel Wedad Jasmine Elmaghraby, University of Maryland, welmaghr@rhsmith.umd.edu The panel will discuss the academic interview process and do’s and don’ts associated with the job search. In addition to comments by current and former search chairs, time will be provided for questions and answers. 2 - Panelist Volodymyr O Babich, Georgetown University, vob2@georgetown.edu 3 - Panelist Candace Arai Yano, University of California-Berkeley, yano@ieor.berkeley.edu 4 - Panelist Brian Tomlin, Tuck School of Business, brian.tomlin@tuck.dartmouth.edu 5 - Panelist Ken Moon, Univeristy of Pennsylvania, Wharton School, Philadelphia, PA, 19104, United States, kenmoon@wharton.upenn.edu 6 - Panelist Ken Moon, Univeristy of Pennsylvania, Philadelphia, PA, United States, kenmoon@wharton.upenn.edu TA16 105A-MCC Energy Systems Sponsored: Optimization, Optimization Under Uncertainty Sponsored Session Chair: Bruno Fanzeres, Visiting PhD Student, Georgia Institute of Technology, 755 Ferst Drive, NW, Atlanta, GA, 30332, United States, santosbruno85@gmail.com 1 - The Information-collecting Vehicle Routing Problem For Emergency Storm Response Lina Al-Kanj, Postdoctoral Research Associate, Princeton University, Olden Street, Sherrerd Hall, room 119, Princeton, NJ, 08544, United States, lalkanj@princeton.edu, Warren B Powell” This talk presents a new policy that routes a utility truck to restore outages in the power grid using trouble calls and the truck’s route as a mechanism for collecting information to create beliefs about outages. This means that routing decisions change our belief about the network, creating the first stochastic vehicle routing problem that explicitly models information collection. The problem is formulated as a sequential stochastic optimization program. Then, a stochastic lookahead policy is presented that uses Monte Carlo tree search (MCTS) to produce a practical policy that is asymptotically optimal. Simulation results show that the developed policy has a close-to-optimal performance.

237