Basic HTML Version
Poster Full Abstracts - Evolution and Quantitative Genetics
Poster board number is above title. The first author is the presenter
265
mechanism and have organized them into a new database. This database will prove useful to researchers in validating or refuting each proposed mechanism
of intron gain and loss.
497B
Statistical models for RNA-seq data.
Rhonda L. Bacher
1
, Justin Dalton
2
, Rita M. Graze
3
, Kurt Jensen
4
, Jonatan Sanchez-Garcia
4
, Pedro Fernandez-Funez
4
,
Diego E. Rincon-Limas
4
, Michelle N. Arbeitman
2
, Ann L. Oberg
5
, Sergey V. Nuzhdin
6
, Lauren M. McIntyre
3
. 1) Departments of Statistics and Mathematics,
University of Florida, Gainesville, Florida, USA; 2) Department of Biomedical Sciences, College of Medicine, Florida State University, Tallahassee,
Florida, USA; 3) Department of Molecular Genetics and Microbiology, University of Florida, Gainesville, Florida, USA; 4) Department of Neurology,
McKnight Brain Institute, University of Florida, Gainesville, Florida, USA; 5) Department of Health Sciences Research, Division of Biomedical Statistics
and Informatics, Mayo Clinic, Rochester, Minnesota, USA; 6) Molecular and Computational Biology, Department of Biological Sciences, University of
Southern California, Los Angeles, California, USA.
RNA-Seq is a tool used for assessing gene expression based on read counts from high throughput sequencing. Many analysis methods to detect differential
expression thus far have focused on using the Poisson distribution, or the negative binomial distribution for analysis of differential expression. Looking at
both the underlying data and the measurement on which we perform the analysis, we propose that RNA-Seq data can be considered continuous. The
underlying libraries are made from a solution of mRNA quantified by concentration. This solution is sampled and sequencing technology is used to estimate
the number of molecules in the sample for a particular gene. Normalization techniques, which result in non-integer values, are often applied to RNA-Seq
data in order to account for systematic effects on the total number of counts, for example the length of the exon/transcript and the total number of reads
mapped to the reference per sample. Furthermore, the raw read counts themselves take on a large range of values (0, 1, to 6 and 7 digit numbers). We
examine three different experiments in drosophila and using a mixed effects model with a normal distribution, we find the residuals do conform to
underlying assumptions when data are complete. When some observations are missing residual assumptions are often violated.
498C
Laboratory selection on Drosophila melanogaster using Bacillus cereus spores: direct response to selection and correlated life history trait
responses.
Lawrence Harshman, Junjie Ma, Andrew Benson, Stephen Kachman, Zhen Hu. Univ Nebraska - Lincoln, Lincoln, NE.
D. melanogaster is a model for laboratory selection experiments. We selected for adult fly survival after tungsten needle-mediated infection with spores
from the bacterium, Bacillus cereus, a gram-positive species closely related to Bacillus anthracis. There were nine lines (populations): three lines were
selected for survival after spore infection, in addition there were three sham-infected control lines and three control lines that were not infected or poked.
After 15 - 20 generations of selection, a strong response was observed for infection survival with > 10-fold increase in the LD50 value for live spores. All
lines were assayed for correlated responses in life history traits after inoculation with autoclaved spores, inoculation with water, or no inoculation. Life span
differences among the lines were observed only when autoclaved spores were introduced. In terms of fecundity, selected lines exhibited considerably higher
levels of life-time egg and progeny production than the control lines. Development time (egg-to-adult) was also investigated. Males and females from the
selected and control lines where separately subjected to three environmental conditions and thus there was a complex matrix of outcomes. One result was
that when selected line males were exposed to autoclaved spores their progeny developed more rapidly. However, when selected line females were exposed
to autoclaved spores their offspring developed more slowly. Two results are potentially noteworthy from an evolutionary standpoint. The first is that there
was no trade-off between evolved survival after spore infection and progeny production. Instead elevated levels of egg and progeny output apparently
compensate for the negative effects of spores on fecundity. The second is the observation of contrasting maternal and paternal effects on progeny
development time in the selected lines.
499A
Computational modeling of
cis
-regulatory modules from 3D expression data in a
Drosophila
blastoderm atlas.
Soile V E Keränen, Oliver Rübel, Mark
D Biggin, David W Knowles. Lawrence Berkeley Natl Lab, Berkeley, CA.
Animal
cis
-regulatory modules (CRMs) function in the 3D context of the whole organism. To generate correct spatial outputs, CRMs must exploit the
spatial information in expression patterns of
trans
-regulatory proteins. Based on the assumption that CRMs are optimized to interpret spatial information, we
used computational methods to evolve ‘CRMs' that generate
in silico
3D spatial expression patterns that match selected 3D target patterns as closely as the
simulation conditions allow. As input data, we used 3D quantitative expression patterns recorded in a VirtualEmbryo; a cellular resolution morphology and
expression atlas of
Drosophila melanogaster
late blastoderm generated by the Berkeley
Drosophila
Transcription Network Project (BDNTP)
(http://bdtnp.lbl.gov/Fly-Net/). Our algorithm used a simple mutation-selection approach to evolve black-box 'CRMs' that could take 3D expression data for
17 transcription factors and use it to generate a 3D output pattern resembling one of the
in vivo
target patterns. To test for convergence and parallelism, we
repeated the evolution 50 times for each target pattern. Overall, normalized, most of the evolved interactions were weak or moderate whereas only few were
strong, but even quite weak interaction could have a selectable effect on output pattern. This behavior is in agreement with the Continuous Network model
proposed based on ChIP-seq data in which CRMs are bound and regulated by large numbers of transcription factors, many of which have only modest
quantitative effects. Moreover, for a given target pattern and a set of transcription factors the solutions often tend to resemble each other, indicating the
existence of a favored solution. This phenomenon may have implications on convergence and parallelism also
in vivo
.
500B
RNA-seq: the challenge.
Lauren M. McIntyre
1
, Rita Graze
1
, Luis Novello
2
, George Casella
2
, Kenny Lopiano
2
, Linda Young
2
, Ann Oberg
3
, Sergey V.
Nuzhdin
4
. 1) Dept Molec Gen & Micro, Univ Florida, Gainesville, FL; 2) Dept Statistics, Univ Florida, Gainesville, FL; 3) Mayo Clinic Rochester, MN; 4)
University of Southern California.
RNA-seq is revolutionizing the way we study transcriptomes. mRNA can be surveyed without prior knowledge of gene transcripts. Alternative splicing of
transcript isoforms and allele specific expression are two applications of this new technology that are particularly exciting. Reports of differences in exon
usage, and splicing between samples as well as differences among alleles and the best way to model and quantify these differences are a subject of great
interest. This new technology has novel challenges. Some challenges are bioinformatics (e.g. map bias); some are technical (e.g. lane to lane variability), and