8:45-9:00 Jacalyn M. Huband, University of West Florida
Teen Sexuality and Cluster Analysis: What's the Relation?
A team of sociologists from The Ohio State University recently surveyed the students at a "typical" American high school to determine the underlying structure of the students' romantic and sexual network. The resulting data shows that approximately 35% of the high school's population can be linked in one large network. Although the largest cluster grabs our attention (and that of the news media), is it the best representation of the data? Or, can we learn more about the network structure (and substructure) through cluster analysis? In this talk, we illustrate the cluster tendency of the relational data, as determined by the VAT (Visual Assessment of Cluster Tendency) algorithm.
9:05-9:20 Oluyede Broderick, Georgia Southern University
Some General Results on Variability Orderings via Probability Weighted Moments and Vitality Distribution
Some general notions of variability orderings are introduced via probability weighted moments and vitality measures. The notions of probability weighted dominance, increasing (decreasing) probability of sudden death (IPSD) and vitality order are defined and explored. Applications to ordering of distribution functions or related random variables, reliability measures and sequences of random variables are presented.
9:25-9:40 William Seffens, Clark Atlanta University
Whole Transcriptome Variance Analysis of mRNA Secondary Structure Free Energies in Human, Mouse and Arabidopsis
Several studies have demonstrated that mRNA stability is an important factor in gene expression. A previous examination of 51 random mRNAs revealed a bias toward more negative folding energies in native sequences, controlling for mononucleotide composition, encoded amino acid sequence and codon usage. Because RNA folding free energies depend on dinucleotide composition, it is a factor that must be controlled for in order to confidently assess whether there is evidence of selection for or against RNA structure in any set of sequences. Instead of examining a limited number of genes, this study considered a complete set of human mRNAs termed a transcriptome and analyzed the variation of energy values among genes of differing length.
Human mRNA sequences were extracted from a reference set called “rna.gbk.gz” obtained from NIH/NCBI, and folding free energies calculated using RNAstructure. Statistical analysis (ANOVA) between the native mRNA and its randomized sequences, and comparing all mRNAs in the human transcriptome, found that the native mRNAs were more stable (greater negative free energy of folding). These studies have shown a highly significant and widespread bias toward local secondary structure.
About 6000 Human, 12000 Mouse and 17000 arabidopsis genes with mRNAs of length 3000 or less base pairs were investigated in detail. For every length L, the native free energy is normally distributed with mean 30.714-0.281*L for Arabidopsis, 3.577-0.317*L for Mouse and 2.75-0.320*L for Human. Variances of 0.02*L, 0.039*L and 0.054*L respectively were calculated for the three transcriptomes. The Human data had many more outliers than Mouse or Arabidopsis. This suggests a possible explanation for why the number of genes is similar across genomes of varying complexity. The human transcriptsome may have more secondary structure to support more complicated gene regulation mechanisms. The randomized sequence folding free energies were found to have the same distribution but with means having different slopes for the three transcriptomes
9:45-10:00
Gary W. Huband, Modeling and Simulation Section, 96 Communications Group, Eglin AFB, FL
Curve Fitting Measured Mass Properties for Small Aircraft
The Air Force is currently developing a new target drone to replace the current dwindling inventory of aging target drones. During flight testing a high fidelity, six degree of freedom simulation of the drone is used to verify the validity and safety of test points, to check that all test points can be accomplished with the available fuel, and to train the drone controllers. Accurate mass properties (center of gravity location and moments of inertia) versus fuel volume are necessary to model drone dynamics. The mass properties of the drone are first estimated from the design then measured after the first few drones are delivered. Because of cost and time constraints only a few points are collected for each property and curve fitting is used to provide mass property versus fuel equations for the simulation. The original curve fits were performed with polynomials which gave good mathematical fits, but resulted in unphysical behavior in certain regions. This presentation will explore curve fitting the data using rational functions based on the definition of the mass property. It will be shown that the rational function curve fits accurately model the mass properties and provide insight into measurement errors and ways of improving the measurements.