Term
|
Definition
*2 traits that have correlated pheno
*can use genetic correlation to reach breeding objective faster |
|
|
Term
|
Definition
+ CORRELATION
*COV(Pyw,Paw) ≠ 0 (+ number)
*EX: inc on trait, inc a different trait
- CORRELATION (INVERSELY CORRELATED)
*inc in one trait dec another
*EX: MY and %fat |
|
|
Term
| what causes traits to be correlated |
|
Definition
GENETICS
*genetics that predispose an ind to grow faster would cause them to have high avg yw ad aw
ENV
*correlated env effects (same env effect may influence both yw and aw)
*EX: an ind may eat more when young, but also eat more when old
COMBINATION |
|
|
Term
|
Definition
COV(Pya,Paw) = COV(BVyw,BVaw) + COV(Eaw,Eyw)
COV(BVyw,BVaw): genetic correlation
COV(Eaw,Eyw): environmental correlation
*genetics and env do not covary ==> so pop w/ better genetics doesnt necessarily mean better E
*the extent to which pheno values are correlated depends on corr of genetics and E |
|
|
Term
| How can quantify the response to selction for T2 give T1 |
|
Definition
Rt1 = [rBVt1,Pt2] * σBVt1 * i
rBVt1,Pt2 = [rBVt1,BVt2 * rBVt2Pt2]
Rt1 = [rBVt1,BVt2 * rBVt2Pt2] * σBVt1 * i
*the corr betwee BV @ T1 and P @ T2 is the product of the correlation between BVt1 and BVt2 time the corr betwee BVt2 and Pt2
*if we are performing pheno selction @ T2 (Pt2), the Rt1 depends on how corr genetics are between traits and the extent to which Pt2 predict BVt2 and amoutn of genetic variationT1 and intensity |
|
|
Term
|
Definition
DIRECT SELECTION (SC = Pt1)
DRt1 = rBVt1,Pt1 * σBVt1 * i = ht1 *σBVt1 * i
DR: direct response; select on Pt1 to predict ^BV for t1
INDIRECT SELECTION (CORRELATED SELECTION) (SC = Pt2)
IRt1 = rBVt1,Pt2 * σBVt1 * i = rBVt1,BVt2 * ht2 * σBVt1 * i
IR: indirect response; what is resp in T1 when selection on Pt2 |
|
|
Term
| What determines the direct response on a trait |
|
Definition
*strength of relationship between BVs and Ps for that trait (h)
*the amount of gentic variation for that trait (σBV)
*intensity (i) |
|
|
Term
|
Definition
| DRt1 / IRt1 = ht1 / (rBVt1,BVt2 * ht2) |
|
|
Term
|
Definition
*is rBVt1,BVt2 * ht2 > ht1 then indirect slection has greater R
*if t1 has low h2, direct selection not effective (b/c Pt1 not good indic of BV)
*if T2 has high h2, BVt2 good pred of BVt1, so can use Pt2 to determine BVt1 (could be better than using Pt1 to determine BVt1)
-if genetic corr (relation bet BVt1 and BVt2) and T2 has high h2 --> looking at Pt2 can tell BVt1
*should pay attention to indirect seletion bc could possibly get better R |
|
|
Term
| If traits are correlated, will always get an IR |
|
Definition
*pheno corr caused by genetics and E
*if traits are corr ony due to E, then rBVt1,BVt2 = 0 (no corr betwee BV of both tratits) and there is NO IR
*just b/c Ps are corr, doesnt mean gentic corr
-need gentic corr to have IR |
|
|
Term
| If IR doesnt give better R, why can you still use it |
|
Definition
1) TRAITS HARD TO MEASURE
*EX: traits that come late in an ind life
*can use a corr trait that can be measured easier/earlier --> tell about other trait --> perform selection in convenient way
2) SEXUAL DIMORPHISM
*EX: MY in bulls
*if could measure a corr trait in a bull --> look @ P corr trait tells us about genetic of trait cant be measured |
|
|
Term
| Ways to categorize traits |
|
Definition
1) PHENO MANIFESTATION
*categorical- phenos in categories (EX: coat color)
*quantitative- phenos described in quant scale w/ cont distrib (EX: ht)
2) GENETIC NATURE
*SIT- affected by sm # genes
*Polygenic- affected by many genes |
|
|
Term
|
Definition
*3rd class of traits (SIT, quant, thresh)
*traits that are polygenic in nature but pheno is manifested categorically
*EX: diseases =- have many loci influence susceptibility |
|
|
Term
| What freq of offspring will be diseased |
|
Definition
USE:
p = M + G + E
*NOTE: cant use HWE b/c not SIT (threshold traits are polygenic) |
|
|
Term
| Distribution in quantitative traits |
|
Definition
*ind w/ pheno values near mean are more freq than extremes (b/c there are many ways to get intermediate pheno from allee combos)
*quant traits have OBSERVABLE PHENOTYPIC DISTRIB ==> can directly observe and understand distrib |
|
|
Term
| Distribution in threshold traits |
|
Definition
*threshold traits have OBSERVABLE PHENO CATEGORIES
*assume unobservable liability distrib
-doesnt have cont distrib: instead assume ind in pop have cont distrib of liability values
-if liability above or below threshold means have diff categories
*cant see norm distrib but know it exists
L < T
*PHENO = CAT A
L ≥ T
*PHENO = CAT B |
|
|
Term
| What does threshold distrib mean |
|
Definition
*we can assume liability exists b/c an ind in same category has same pheno, but genetics predisposes them differently on the scale (pheno manifestation is determined by threshold and LV)
*can assume norm distrib b/c polygenic traits prod normal distrib b/c have many allele combos that predisposes ind to be close to mean, and few further
*shows that 2 ind w/ same pheno can have diff genetics |
|
|
Term
|
Definition
*genetic predisposition where inds w/in same category can have diff gentics (can also be E)
L = M + BV + GCV + E |
|
|
Term
VISUAL REPRESENTATION:
Where is threshold? |
|
Definition
| *leave ind you are selecting on upper end (right) |
|
|
Term
VISUAL REPRESENTATION:
WHat is the corresponding i for top ___% |
|
Definition
*use table to find i from p(%)
*i always + b/c doesnt have a scale b.c expressed as SD |
|
|
Term
VISUAL REPRESENTATION:
What does i represent for threshold and where to find on axis |
|
Definition
* i is the liability of select ind expressed as SD
*distance between M and selected ind |
|
|
Term
| What is the response to selection for threshold traits? / How much does the mean move in the next gen? |
|
Definition
R = h2 * i
*if GIVEN h2=0.5 ==> means parents transmit only 1/2 libabillity b/c 50% caused by genetics, other 50% is nonadd and E |
|
|
Term
| What is happening in next gen |
|
Definition
*R only 1/2 of mea nliab transfered (i)
-start at mid curve go to mid of i
*in next gen, mean moves up
-new mean is at end of R
*threshold stays in same place, but the proportion of ind in e. category moves |
|
|
Term
| How many ind will be diseased in G2 |
|
Definition
*relation between proportion saved and distance between M and T (x)
*look at table
*solve for p
p = |
|
|
Term
|
Definition
*distance between M and T
*x2 = x1 - R
*solve for p |
|
|
Term
| genomic prediction/ selection |
|
Definition
*using genomic data (i.e whole genome) to predict BVs and perform selection based on those estimates
*need to know genome of species working w/
*slection w/o pheno records |
|
|
Term
| technology for genomic selection |
|
Definition
1) BOVINE GENOME SEQ
2) IDENTIF OF DNA SEQ VARIATION
*SNP (sing nuc polymo) ==> discvoer variation form
refernece genome
3) STATISTIACL METHODS
*to est allelic efefcts of genes affecting QTL (Bayesian)
4) GENOTPYING CHIP
*to geno animals for many markers
*cheap method for lrg # ind and lrg # variants |
|
|
Term
|
Definition
*need to use data to predict BV
1) YOUNG SIRE PARENT AVG
✔: can get immediately
X: dec h (0.2-0.38), Mend samp
*even if know exact BV of parents, Mend samp occurs so dont know exactly whats passed
*using parent avg doesnt account for what alleles are actually trasmitted due to mend samp
2) YOUNG SIRE PROGENY TEST
✔: very high accuracy
X: expensive, takes lots of years
3) YOUNG SIRE GENOMIC SELECTION
✔: can do as soon as born and looks directly @ inherited alleles; accounts for mend samp to give MBV; pretty h, fast, cheap |
|
|
Term
|
Definition
| cell types vary b/c have diff genes expressed in e. type (even though e. cell contains the same genome) |
|
|
Term
|
Definition
TRAINING SET (MODEL DEVELOPMENT)
*known EBV and pheno for pop A1
*estimate relation between true BV and geno to est MBV
*form a regression
CROSS VALIDATE (MODEL OPTIMIZATION)
*known EBV and pheno for pop A2
*use #1 and apply to ind w/ known BVs to predict
*test how good relation between BV and geno to est MBV
* form a corr (test h)
---
PREDICTION (MODEL IMPLEMENTATION)
*unknown EBV and pheno
*predict MBV in pop B
*if #1&2 work, can apply to ind w/ unknown BV |
|
|
Term
| How to estimate the effect of e. SNP on EBV |
|
Definition
*use a regression moel
*estimate the average effetc of an allele substitution (a) |
|
|
Term
|
Definition
*molecular breeding value
*BV predicted from adding alleleic effects across all loci |
|
|
Term
|
Definition
PROVIDES FASTER BUT EQUIVALENT INFO
*having genotyping done on a calf is equivalent to waiting 5yrs for the ind to have 34 daughters
INCREASE VALUE IN TRAITS W/ LOW h2
*if h2 low, the geno is more valuable ==> 1SNP = 131 daughters
1) inc h
*b/c more info quickly
2) dec L
*b/c can use young ind more often b/c info quick and reliable
3) inc i
*b/c young ind are more accurate so can keep a sm number
NO TRADEOFFS |
|
|
Term
| expected response to selction |
|
Definition
*genetic gain per unit time
^BV / t = h2 * op * i / L |
|
|
Term
|
Definition
*can use genomics to look at inbreeding
0- homo for an alles
1- hetero
2- homo for other allele
*can tell an ind is extremely inbred if has low # heteros
(low homos = more inbred) |
|
|
Term
| Impact of genomic selection |
|
Definition
1) inc # animals being testested and useed as breedrs
2) reduction in ibredding/managemtn - percentage testing
3) accurate estimation of BV of animals w/ little or no pheno records
4) elimination of age and sex limits in geneitc improvemtn
*can geno immediately on sex dimorphic animals
5)Specialized mating by genotyping dams
*complimenatrity mating, man/min heterozyg
6)inc i and h
7) dec L |
|
|
