19th Twin and Family Studies Methodology Workshop Report by jwc45040


									Abdullah Sheikhi
Curtin University of Technology                               19th Twin Workshop Report

          19th Twin and Family Studies Methodology Workshop Report
                          Boulder 6th – 10 March 2006

Q: Submit a brief report (1-2 pages) detailing what is being learned or gained
from attendance this workshop?

The sessions I had from the workshop were enriched by various topics in the area of
behavioural genetics and many of new skills have being learned from it.

In the talk presented by Nick Martin about the Causes of Human variation I learned
the different approaches about the analysis of the causes of human variability and the
idea of variance which was developed by Fisher (1918). Also I had a historical
overview about the pioneer people in this field from Darwin and Mendel in 1860s
until 2000 by Michael Neal. Also I had an overview about the complex trait model
and how the variance components can interact with each other. Furthermore, I had an
idea about the designs to disentangle Genetic + Environment. Such as family studies,
MZ twins alone, MZ twins reared apart, adoptions, MZ and DZ twins reared together,
and extended twin design. Moreover, I had an introduction about the ACE Model for
twin data and structural equation modeling.

In the second session presented by Mike Neale and Hermine Maese titled
“Summarizing Variation & Matrix algebra/ Introduction to Mx” I had the opportunity
to learn more about the Mean, Variance, and Covariance and how can be calculated
and how to estimate these parameter by Maximum Likelihood. Also I refreshed my
memory about the use of Matrix Algebra and Normal Likelihood Theory. Also, I had
an overview about how to get used to Mx script language such as the general
comments, Job Structure, Group Structure, Read Observed Data, Matrix Declaration,
Matrix Types: Matrices II and III, Constrained Matrices, Binary Matrix Operations,
Matrix Operations Priorities, Matrix Functions, Specify Numbers/ Parameters.

In the third talk by Danielle Dick gave the concept of Path Analysis which showed me
how to represent linear models for the relationships between variables in
diagrammatic form, and explained how path analysis can make it easy to derive
expectation for the variances and covariances of variables in terms of the parameters
proposed by the model. Also she showed how much path analysis can be easily
translated into matrix form for use in programs such as Mx.

The forth session titled “Univariate ACE Model” presented by Hermine Maes and
Nick Martin, reviewed the basic elements of doing univariate genetic analysis. I have
learned the difference between saturated models and Univariate Models and also the
components of ACE and ADE models. Also, I had the opportunity to apply both
models on actual data by estimating the variance components. In addition, I got my
self familiar with the Mx script structures and commands.

In the fifth presentation titled “Biometrical Genetics” given by Manuel Ferreira, I had
the chance to know how to use the common genetic parameters such as allele
frequencies, genetic effects, dominance, variance components, etc. in order to
construct biometrical genetic model.
Abdullah Sheikhi
Curtin University of Technology                              19th Twin Workshop Report

 In the sixth talk titled “Categorical Data“presented by Fruhling Rijsdijk and Kate
Morley, I had an introduction about the Categorical Data and how can define liability
and describe assumptions of the liability model. Also, I gained the skill about how
heritability of liability can be estimated from categorical twin data. Furthermore, I
learned how can I fit ordinal data in Mx as Mx has a built-in fit function for the
maximum-likelihood analysis of 2-way Contingency tables analyses limited to only
two variables.

The seventh and eightieth sessions were talked about Heterogeneity presented by
Hermine Maes and Danielle Posthuma, gave me a background about the
contributions of additive genetic, dominance/shared environmental and unique
environmental factors to the variance for both the Univariate Analysis and
Heterogeneity Analysis, and also if there are different genetic/environmental factors
influencing the trait in males and females.

The ninth talk was about the Missing Data theory by Mike Neale. In this presentation
I had the opportunity to know types of missing data and I learned the way to structure
the missing data in Mx script

The tenth presentation by Danielle Posthuma titled “Extended sibshihps”, explained
to us the need to design an extended model of MZ or DZ twin with large sibships. I
had the chance to practice how to include up to six sibs to the MZ and DZ twin
pedigree, and checked the –2ll, df, estimated parameters, n observations for each
model in Mx script. In addition, I exercised how to modify the Mx script for
maximum nr of siblings = 3, 4, 5, or 6, in both univariate and bivariate models.

The eleventh and twelfth presentations were by Manuel Ferreira and Sarah Medland
titled „Introduction to the Principles of QTL Linkage analysis in Mx‟, which helps to
localise and identify a locus that regulates a trait (QTL). From this talk I had the
opportunity to strengthen my background about the difference between Parametric vs.
Nonparametric Linkage Analysis, and to develop the concept of the two kinds of
genotypic similarity between relatives; the IBS Alleles shared Identical By State
“look the same”, and IBD Alleles shared Identical By Descent “a copy of the same
ancestor allele”. In addition, I increased my knowledge about the Haseman-Elston
regression for the Quantitative traits, Variance Components Maximum Likelihood for
the Quantitative & Categorical traits, and genome-wide linkage analysis. I also had
the opportunity to strengthen my skills on MERLIN to estimate IBD and allele
frequency for a pedigree.

In the thirteenth talk by Benjamin Neals titled „Association Analysis‟ I had a wide
spectrum view about the strategies of association and linkage disequilibrium designs,
including: linear Model of Association, types of Allelic Association, building
haplotype maps (blocks) for gene-finding, biometrical model, regression model of
association, family-based - Trio (TDT) design, sib-pairs/extended families (QTDT)
design, and case-control design. I also had an overview about a wide range of
statistical methods that applied with association from t-test to evolutionary model-
based MCMC in order to correlate phenotypic and genotypic variability.
Abdullah Sheikhi
Curtin University of Technology                              19th Twin Workshop Report

The final talk was by Nick Martin titled „QTL studies: past, present and future‟. This
presentation exhibited the significance role of the of the previous twin genetic
strategies including AEC model by Mx, linkage and association. This had been
exhibited by several MZ and DZ twin genetic studies on complex disorders such as
Epilepsy, Schizophrenia, Depression and Asthma. Additionally, he talked about the
effect of multivariate analysis on linkage power, and epigenetic discordance in
identical twins.

In conclusion, this workshop has acquired me the fundamentals about the phenotypic
approaches on twins by using AEC univariate and multivariate twin designs by using
Mx. Moreover, I have learned the basis of Linkage and association analyses that help
to localise and identify genes that contributed to certain phenotypes including the
human complex disorders.

 Finally, I would like to introduce my sincere thanks to the Australian Twin Registry
for their funding to attend this workshop, which has helped me to gain new skills and
knowledge on twin research.

To top