This experiment was designed originally to sample various meat and carcass quality
aspects of Ontario pigs from 1990 to 1993. It wasn’t designed to detect QTL associated
with meat quality characteristics of pork, but we had enough meat samples remaining
from this experiment to have DNA that we could add this component. The population was an outbred population.
Moisture retained by meat has a connection with the flavour and juiciness of the cooked
product. The results below are from an analysis of drip loss (moisture lost from a fresh
piece of pork in 24 hours) and genotypes derived from the meat samples.
In this population, the polymorphism that was detected was labelled MspI (which is a pituitary transcription factor
gene), and the genotypes were labelled 1 = A1A1, 2 = A1A2 and 3 = A2A2. The allele frequencies were
f(A1) = 0.68 and f(A2) = 0.32. Although the genotypic frequencies were not exactly p2,
2pq and q2, for the purposes of this lab, you can assume they were. The model for this
analysis included hot carcass weight (HOTCWT1) to adjust for carcass size differences,
the test barn fill number (FILLNO) to account for known environmental factors, sex
(males, females and barrows – castrated males). Drip loss is measured in percentage and
less drip loss is better (more moisture retained in the meat).
Some important information is contained in the printout below.
This is an output from SAS. The statistical tests in SAS are all accompanied by
probability values (probability of a larger F value, "Pr > F" and larger t value, "Pr > |t|").
This is a p-value and to make a statistical test, you compare these probabilities to the
critical value you have established for the experiment. If the Pr > F is less than your
critical value (I recommend using 0.05 as your critical value) then the factor being tested is significant.
The Pr > |t| column tests the difference of means from zero so factors with this column are being
tested as significantly different from zero. “LSMeans” refers to an adjusted mean (average) that is
used as a measure of the average of each factor in the model, including the genotypes. As
noted above, the Pr > |t| column tests the difference of each LSMean from zero.
The GLM Procedure
Dependent Variable: DLLNPCT which is Loin Drip Loss %
Sum of
Source DF Squares Mean Square F Value Pr > F
Model 36 1564.325289 43.453480 3.53 <.0001
Error 211 2593.732982 12.292573
Corrected Total 247 4158.058271
R-Square Coeff Var Root MSE DLLNPCT Mean
0.376215 29.25578 3.506077 11.98422
Source DF Type III SS Mean Square F Value Pr > F
HOTCWT1 1 32.8255304 32.8255304 2.67 0.1037
FILLNO 22 714.3077354 32.4685334 2.64 0.0002
SEX 2 47.9965480 23.9982740 1.95 0.1445
MspI 2 189.9263802 94.9631901 7.73 0.0006
Standard
Parameter Estimate Error t Value Pr > |t|
Intercept 7.223163310 B 5.42727411 1.33 0.1847
HOTCWT1 0.075560897 0.04623945 1.63 0.1037
SEX B 1.136799775 B 0.61026699 1.86 0.0639
SEX F 0.841730577 B 0.56550265 1.49 0.1381
SEX M 0.000000000 B . . .
MspI 1 -3.836074399 B 1.23681260 -3.10 0.0022
MspI 2 -2.910062857 B 1.09289797 -3.49 0.0006
MspI 3 0.000000000 B . . .
Least Squares Means
MspI DLLNPCT Standard
LSMEAN Error Pr > |t|
1 13.2747909 0.5631372 <.0001
2 10.3647280 1.0321316 <.0001
3 9.4387165 1.1460667 <.0001
Refer to the SAS output sample from question 1 above. Based on a
t-test of the difference of a value from zero, which parameters
were statistically significantly different from zero?
Question 2 options:
|
|
|
2)
|
Sex B (Barrows) and Sex F (Females) |
|
|
3)
|
Intercept and hot carcass weight |
|
|
4)
|
MSP1 genotypes 1 and 2 |
|
|
|