Read Me! Grisel JEGrisel JE. Quantitative trait locus analysis.Alcohol Res Health. 2000;24(3):169-74.

Slides:



Advertisements
Similar presentations
Introduction Materials and methods SUBJECTS : Balb/cJ and C57BL/6J inbred mouse strains, and inbred fruit fly strains number 11 and 70 from the recombinant.
Advertisements

Experimental crosses. Inbred Strain Cross Backcross.
Qualitative and Quantitative traits
Lecture 3 HSPM J716. Efficiency in an estimator Efficiency = low bias and low variance Unbiased with high variance – not very useful Biased with low variance.
FTP Biostatistics II Model parameter estimations: Confronting models with measurements.
QTL Mapping R. M. Sundaram.
1 QTL mapping in mice Lecture 10, Statistics 246 February 24, 2004.
Quantitative Genetics Theoretical justification Estimation of heritability –Family studies –Response to selection –Inbred strain comparisons Quantitative.
MULTIPLE REGRESSION. OVERVIEW What Makes it Multiple? What Makes it Multiple? Additional Assumptions Additional Assumptions Methods of Entering Variables.
LINEAR REGRESSION: Evaluating Regression Models. Overview Standard Error of the Estimate Goodness of Fit Coefficient of Determination Regression Coefficients.
Regression and Correlation
Simple Regression correlation vs. prediction research prediction and relationship strength interpreting regression formulas –quantitative vs. binary predictor.
1 MF-852 Financial Econometrics Lecture 6 Linear Regression I Roy J. Epstein Fall 2003.
1 4. Multiple Regression I ECON 251 Research Methods.
Today Concepts underlying inferential statistics
Experiments in Plant Hybridization (1865) by Gregor Mendel Bad title! People forgot about me and my work!
C82MCP Diploma Statistics School of Psychology University of Nottingham 1 Linear Regression and Linear Prediction Predicting the score on one variable.
Quantitative Genetics
Elec471 Embedded Computer Systems Chapter 4, Probability and Statistics By Prof. Tim Johnson, PE Wentworth Institute of Technology Boston, MA Theory and.
Issues in Experimental Design Reliability and ‘Error’
Psy B07 Chapter 1Slide 1 ANALYSIS OF VARIANCE. Psy B07 Chapter 1Slide 2 t-test refresher  In chapter 7 we talked about analyses that could be conducted.
Chapter 4 Hypothesis Testing, Power, and Control: A Review of the Basics.
Lecture 5: Segregation Analysis I Date: 9/10/02  Counting number of genotypes, mating types  Segregation analysis: dominant, codominant, estimating segregation.
David Corne, and Nick Taylor, Heriot-Watt University - These slides and related resources:
Fall 2013 Lecture 5: Chapter 5 Statistical Analysis of Data …yes the “S” word.
Chapter 9 Comparing More than Two Means. Review of Simulation-Based Tests  One proportion:  We created a null distribution by flipping a coin, rolling.
CORRELATION & REGRESSION
Biostatistics Unit 9 – Regression and Correlation.
Testing Theories: Three Reasons Why Data Might not Match the Theory Psych 437.
Section Copyright © 2014, 2012, 2010 Pearson Education, Inc. Lecture Slides Elementary Statistics Twelfth Edition and the Triola Statistics Series.
Statistical Power 1. First: Effect Size The size of the distance between two means in standardized units (not inferential). A measure of the impact of.
User Study Evaluation Human-Computer Interaction.
Section Copyright © 2014, 2012, 2010 Pearson Education, Inc. Lecture Slides Elementary Statistics Twelfth Edition and the Triola Statistics Series.
Biostatistics in Practice Peter D. Christenson Biostatistician LABioMed.org /Biostat Session 6: Case Study.
Standard Error and Confidence Intervals Martin Bland Professor of Health Statistics University of York
Multiple Linear Regression. Purpose To analyze the relationship between a single dependent variable and several independent variables.
Lecture 5: Chapter 5: Part I: pg Statistical Analysis of Data …yes the “S” word.
MGS3100_04.ppt/Sep 29, 2015/Page 1 Georgia State University - Confidential MGS 3100 Business Analysis Regression Sep 29 and 30, 2015.
PSY2004 Research Methods PSY2005 Applied Research Methods Week Five.
Experimental Design and Data Structure Supplement to Lecture 8 Fall
Quantitative Genetics. Continuous phenotypic variation within populations- not discrete characters Phenotypic variation due to both genetic and environmental.
Complex Traits Most neurobehavioral traits are complex Multifactorial
Quantitative Genetics
Correlation Assume you have two measurements, x and y, on a set of objects, and would like to know if x and y are related. If they are directly related,
Three Statistical Issues (1) Observational Study (2) Multiple Comparisons (3) Censoring Definitions.
VI. Regression Analysis A. Simple Linear Regression 1. Scatter Plots Regression analysis is best taught via an example. Pencil lead is a ceramic material.
Stat 112 Notes 9 Today: –Multicollinearity (Chapter 4.6) –Multiple regression and causal inference.
Education 793 Class Notes Decisions, Error and Power Presentation 8.
Heads Up! Sept 22 – Oct 4 Probability Perceived by many as a difficult topic Get ready ahead of time.
1 Psych 5510/6510 Chapter 14 Repeated Measures ANOVA: Models with Nonindependent Errors Part 1 (Crossed Designs) Spring, 2009.
Lecture 21: Quantitative Traits I Date: 11/05/02  Review: covariance, regression, etc  Introduction to quantitative genetics.
Dept of Bioenvironmental Systems Engineering National Taiwan University Lab for Remote Sensing Hydrology and Spatial Modeling STATISTICS Linear Statistical.
1 Estimation of Gene-Specific Variance 2/17/2011 Copyright © 2011 Dan Nettleton.
Example x y We wish to check for a non zero correlation.
Pedagogical Objectives Bioinformatics/Neuroinformatics Unit Review of genetics Review/introduction of statistical analyses and concepts Introduce QTL.
Statistical Analysis An Introduction to MRI Physics and Analysis Michael Jay Schillaci, PhD Monday, April 7 th, 2007.
Lecture 22: Quantitative Traits II
Lecture 23: Quantitative Traits III Date: 11/12/02  Single locus backcross regression  Single locus backcross likelihood  F2 – regression, likelihood,
Chapter 22 - Quantitative genetics: Traits with a continuous distribution of phenotypes are called continuous traits (e.g., height, weight, growth rate,
Marshall University School of Medicine Department of Biochemistry and Microbiology BMS 617 Lecture 6 –Multiple hypothesis testing Marshall University Genomics.
Why you should know about experimental crosses. To save you from embarrassment.
Using Merlin in Rheumatoid Arthritis Analyses Wei V. Chen 05/05/2004.
Regression Analysis: A statistical procedure used to find relations among a set of variables B. Klinkenberg G
The heterozygosity that appears
Genome Wide Association Studies using SNP
Simple Linear Regression
Detecting variance-controlling QTL
MGS 3100 Business Analysis Regression Feb 18, 2016
F test for Lack of Fit The lack of fit test..
Presentation transcript:

Read Me! Grisel JEGrisel JE. Quantitative trait locus analysis.Alcohol Res Health. 2000;24(3):

Read me!

Bioinformatics/Neuroinformatics Unit—Specific steps Quantify phenotype—olfactory bulb volume Remove error variance –Due to differential shrinkage of brains –Due to multiple raters –Due to extraneous variables--demographic characteristics (eg. Sex, age, body weight, brain weight) and other individual differences.

Now that the phenotype has been quantified, we need to clean up these data! First of all, who shrank the brains?

OK--so we have to correct for shrinkage in case it is variable among brains, which it is.

We can correct for the shrinkage if we know the density of brain (1.05 mg/mm 3 ) because we know how much the brain weighed before processing. We can figure out its original volume, then correct for shrinkage of the olfactory bulbs using this formula:

Bioinformatics/Neuroinformatics Unit—Specific steps Quantify phenotype—olfactory bulb volume Remove error variance –Due to differential shrinkage of brains –Due to multiple raters –Due to extraneous variables--demographic characteristics (eg. Sex, age, body weight, brain weight) and other individual differences.

We are getting rid of error due to multiple raters by: 1)using multiple measurers for each mouse--and insisting that these raters agree. 2) using the median value of all measurers of a given mouse.

Bioinformatics/Neuroinformatics Unit—Specific steps Quantify phenotype—olfactory bulb volume Remove error variance –Due to differential shrinkage of brains –Due to multiple raters –Due to extraneous variables--demographic characteristics (eg. Sex, age, body weight, brain weight) and other individual differences.

So, you have quantified the phenotype, Let’s get on with it!

What a mess! Why did you guys use mice of different sexes, ages, body weights, and brain weights? Didn’t your professors ever teach you anything about controls????

HI! I’m Francis Galton, Chuck Darwin’s cousin, and I can help you out of this mess! You need one of my inventions, linear regression, to help you with your lack of control there, Gregor.

By the way, I called it regression, because everybody seemed to regress toward the mean through successive generations!

Obviously, you can’t control for sex, body weight, brain weight, and age at this point! But thanks to me, you can control for these variables by a statistical method--linear regression. Using linear regression allows one to eliminate the variance (differences) in scores associated with these various extraneous variables.

Fortunately, we can assume that variance--statistics talk for the differences among individuals-- is additive.

What is key in using regression to control for various extraneous variables is the additive model of variance.  2 t otal =  2 sex +  2 body weight +   brain weight +  2 age +  2 error +  2 olfactory bulb genes

Thus, the total variance can be partitioned into the variance associated with each of these extraneous variables such as sex, body weight, brain weight, and age. Then we can successively remove the variance associated with each of these variables and hopefully just have residual variance that only pertains to gene effects on olfactory bulbs.

Let us first consider the case of simple linear regression before we tackle the problem of multiple regression.

In regression we predict the y variable from the x.

In regression we predict the y variable from the x.

^ Y Y _ Variance predicted by X Residual (error) Body Weight (grams) OB Volume

The variance left over after the variance from the other variable(s) has been removed is the residual variance. This residual variance is precious to us because it has the variance specific gene effects on olfactory bulbs.

So the SS E SS yy is our treasure, yet another’s trash.

By using multiple regression, We can remove the variance associated with extraneous variables and so statistically control for these variables.

What is key in using regression to control for various extraneous variables is the additive model of variance.  2 t otal =  2 sex +  2 body weight +   brain weight +  2 age +  2 error +  2 olfactory bulb genes

Not yer data! Variables Controlled by Regression

Bioinformatics/Neuroinformatics Unit—Specific steps Quantify phenotype—olfactory bulb volume Remove error variance –Due to differential shrinkage of brains –Due to multiple raters –Due to extraneous variables--demographic characteristics (eg. Sex, age, body weight, brain weight) and other individual differences.

Bioinformatics/Neuroinformatics Unit—Specific steps Quantify phenotype—olfactory bulb volume Remove error variance –Due to differential shrinkage of brains –Due to multiple raters –Due to extraneous variables--demographic characteristics (eg. Sex, age, body weight, brain weight) and other individual differences.

Regression must be done on individuals. After having controlled for various variables by regression, we will average the olfactory bulb values from the various individuals within a given recombinant inbred strain.

Recall, however, that we can think of Each Measurement = Error + True Score  2 error +  2 olfactory bulb genes

If our measures largely error, then relationships with variance in the phenotype with variance in the genome will be watered-down because error tends to work randomly and add noise to our data. Randomness cannot be systematically related to anything.

How to make mistakes with statistics Type II (beta) errors—AKA false negatives –Small effect size –Small n –Greater variance in scores Greater the error variance, the more Type II errors Type I (alpha) errors—false positives –Stringency of the alpha error rate Significant Individual point p = 1.5 x for genome-wide  =.05 Suggested individual point p = 3 x for genome wide  =.63

QTL is good for detecting the approximate locus of multiple genes affecting a phenotype across all the chromosomes, except Y. This is a graph that displays the likelihood ratio statistic as a function of locus on the various chromosomes, which are numbered at top.

Thus, lots of error variance will give us false negatives (Type II errors) when we do QTL analyses!

B D B B B B B B B B B B B B D D D D D D D D D D D D Phenotypic Measurement (Residual) LRS is LOW X all X D X B Marker Type

B D B B B B BB BBBB BB D D D D D D D D D D D D Phenotypic Measurement (Residual) LRS is High X all X D X B Marker Type

B D B B B B B B B B B B B B D D D D D D D D D D D D Phenotypic Measurement (Residual) LRS is LOW X all X D X B Marker Type

We may not replicate Williams et al. for a couple of reasons.

Williams et al. used weight, which is a more objective measure than ours and had fewer observers. Thus, they probably had less error associated with their measures and fewer false negatives (Type II errors).

We excluded all of the anterior olfactory nucleus, whereas Williams et al. (2001) cut through it in an irregular fashion.

In this lecture you have learned about 1) simple and multiple regression and how they can be used to control for extraneous variables, 2) how error is being controlled in your experiment 3) Reasons why you may not have perfectly replicated Williams et al. (2001).

On behalf of your mentors, do enjoy the remains of the day.