Click to edit Master text styles
Genetic similarity can be defined in several ways using:
A= pedigree data.
Wright, who was a
USDA employee btw, used only pedigree data to calculate probabilities that
gene pairs are identical by descent.
G = Genomic relationships use
genotype data to estimate the fraction of total DNA that individuals actually
T = Fractions of alleles that two individuals share for loci that
affect a certain trait will be termed QTL or T relationships.
The term QTL often refers only to loci with
large effects, but here includes all loci that affect a trait.
Inheritance of DNA can be traced within
and across families for hundreds of genes influencing quantitative traits by
genotyping many markers across the genome.
Markers may or may not contain loci that affect a trait. Relationship
matrix of markers….
Vq is the co-variance matrix of QTL’s
U = breeding value or sum of
T = QTL relationship matrix.
Use marker, as not many QTL have been indentified.
Unlikely, unless one gene controls the trait.
Full sibs, genes in common at one
I have named alleles from grandparents,
w x y z.
If wy, and w,y
alleles in common.
If wz, and w,y
have 1, xy, and w,y also 1.
x,z and w,y
0 in common.
Overall 25% have two, 50% have one, 25% have none,
Simple Mendelian genetics.
for one locus,
Through simulations, Paul , and more
recently Mel, has shown that as the
number of independent loci increase, SD decreases.
From 35% for one
locus, 16% for 2 loci, 11% for 10 up until we get to 100. 100 is about the maximum independent loci
our simulations allow, because after that, loci are NOT independent.
if we know everything about Sibs, SD
is zero. But, never get there….. So how related are individuals?
IF pedigrees, stop at 1960,
ancestors are assumed to be unrelated.
But, actually, before the known
somewhere they are related……
A better term may be gametic disequilibrium.
Look at these relationships…….
Hopefully, the previous slides will become clear.
We are all familiar with a traditional
pedigree chart. Animal is expected to
be an average of his parents.
Here is a simple Genomic pedigree
chart. 2 separate chromosome segments
are shown for each ancestor. (During
meiosis , they divide-recombination). The earliest generation, are all one
color for illustrative purposes. Dairy
cattle have 30 chromosomes, including the
X and Y. On average each
chromosome carries millions of base pairs, and thousands of genes.
1)For the first segment,
the sire has inherited ½ from each piece.
In the second segment, the majority was inherited from one piece (both
of these are an example of a crossover).
The individual has received portions of the first segment that are
traced back to the grandparent generation.
2) The dam has inherited most of
one segment from one portion of her sire’s piece, and a mixture (double
crossover) from her dam’s piece. The
individual has segments that can be traced to all grandparents.
In this example…. Say the light green segment from the
paternal grand dam was one that was high milk production; the offspring, inheriting a large segment
from her, could mean that through this recombination, this animal would have a
higher PTA for milk that a full sib.
SNP is a DNA sequence variation that
occurs when a single nucleotide differs between members of a species…
SNP 2 alleles, 2 variants….
Haplotype definition derived from http://en.wikipedia.org/wiki/Haplotype
Very complex, as 100’s of genes, and
1000’s of base pairs…. How to trace
A little better….. Can trace, but still have
difficulties…… so move from a haplotye
to a genomic pedigree…..
One way to do this is to translate
haplotype to genotype.
on how to best to accomplish this.
this example aa is always zero, TT would be 2…..
0 = homozygous for first allele 1 =
heterozygous 2 = homozygous for second allele Many ways to count, this is counting
These numbers could be fed directly into a relationship
matrix, just 0,1,and 2’s.
alleles, minor alleles… frequency, what is major in one breed, possibly not in
In simulations, and Paul will go over
the math in the next presentation. Lots of markers, 1000 QTL’s, not of major
effect, and high rel sibs.
As the number of full sibs increases, in the A matrix, never rise above 50%
reliability. But with a genomic
matix, with 10 full sibs, already above
50%, with 100 77%....
More reliability for a genomic model than a
traditional additive relationship matrix.
Paul follows this talk, and his
presentation will hopefully answer any mathematics questions you may have…