Daly Lab Research Summary
Our
lab is focused on the increasingly important interface
between computational science and genetics. Specifically,
our focus is on statistical genetics and the development
of computational methods and tools for better understanding
complex problems in human genetics.
Understanding
Patterns of Human Genetic Variation. One major
effort in our lab has been to gain a better understanding
of patterns of genetic variation in the human genome
and to translate that knowledge into more effective
statistical methods for finding the variations that
contribute to disease risk and impact individual drug
response. It has been known for some time that most
variation in the genome takes the form of single nucleotide
polymorphisms (SNPs), which are found every few hundred
base pairs throughout the genome. Patterns of alleles
at nearby SNPs were seen to be quite nonrandom, largely
because of limited historical opportunity for recombination.
This knowledge offered the tantalizing hope that human
genetic association studies could become more tractable
and provide researchers with a fuller characterization
of these relationships among polymorphisms. This past
year, we have made tremendous strides in that direction.
In
the course of identifying a risk factor for Crohn’s
disease (in collaboration with J. Rioux, Whitehead Center
for Genome Research), we collected very high density
variation data on hundreds of individuals. With this
data, we were able, for the first time, to characterize
the true nature of human genetic diversity. We discovered
genomic regions of 10 to 100 kilobases in length in
which the vast majority of chromosomes in a population
carried one of only two, three, or four sequences. Furthermore,
we found no evidence of recombination having played
any role in the assortment of sequences in these regions.
However, in between such regions, recombination was
observed to have occurred quite frequently, breaking
and rearranging the common patterns abruptly.
This
finding has led us to propose a new model of human genetic
variation in which recombination is largely confined
to hotspots similiar to those observed in yeast. As
such, the basic unit of genetic variation is not an
individual polymorphic marker, but rather, a genomic
segment, sometimes quite long, located between such
sites of recombination. Encouragingly, this model has
been recently supported by other research groups and
in a large genomic screen of our own, in which this
model appears representative of the vast majority of
the human genome. [in collaboration with D. Altshuler,
Whitehead Center for Genome Research and Massachusetts
General Hospital]
As
a result of this work, the lab has embarked upon several
new projects.The first is the development of a computational
model that properly takes account of the underlying
nature of genetic variation for use in association studies.
We have developed an initial implementation of a novel
method (based on a hidden Markov model) which assigns
individuals to one of the common sequences in a genomic
region and discovers the sites of significant recombination.
This allows us to compare the chromosomal composition
of cases and controls, or responders and non-responders,
in a much more accurate manner than afforded by the
examination of any individual polymorphism. We hope
to enhance and use these methods to discover the elusive
connection between genotype and phenotype. We are also
beginning to pursue detailed examinations of genomic
sequence to discover what signatures and structures
may correlate with hotspots and coldspots of recombination,
as we discover them in the studies of variation.
Linkage
Analysis.We also conduct research in the traditional
statistical genetics field of linkage analysis. We are
continuing to improve and expand the methods implemented
in our GENEHUNTER software, which is used for performing
statistical analyses that identify genomic regions containing
disease risk factors in families. GENEHUNTER is currently
used by hundreds of labs worldwide.
Another
area of interest in our lab is the use of quantitative
measurements as covariates to increase the power of
analyses that are traditionally limited by treating
all cases of a disease as equal—ignoring severity,
age of onset, family history, and relevant quantitative
measurements of disease. We are collaborating with many
other research groups to study a variety of complex
phenotypes. For example, we are working with researchers
in Finland to study asthma and diabetes and with researchers
at Boston’s Children’s Hospital to study
exceptional longevity. We are using this data to test
and refine new statistical methods. We are also working
with international genetic research collaborations that
are studying inflammatory bowel disease and psoriasis,
in order to explore the statistical properties and potential
uses of such studies when results of many related studies
from diverse populations are combined.
|
