Gene expression analysis ANOVA

kubu4 commented 7 years ago

Report results from two-way ANOVA regarding interactions. Was there an interaction between treatment and population?

kubu4 commented 7 years ago

I think this was done, correct? I think it might just need a better explanation in the Statistical Analysis subsection of the METHODS of how the ANOVAs were set up.

sr320 commented 7 years ago

Maybe confer with Brent

kubu4 commented 7 years ago

Brief consultation with Brent suggests that this was indeed done (based on the R script), but was not reported in the manuscript. I'll glance at the ANOVA outputs, see what I can figure out and determine how easy/difficult it will be to add this info to the paper.

kubu4 commented 7 years ago

Here's the output from one ANOVA & Tukey's on CARM. Commands to R are indicated by ">".

I have no idea how to interpret this, so will need someone else to figure out how to answer the reviewer's questions.

> CARM<-aov(CARMlog~Pop+Treat+Pop:Treat, data=dCt)
> CARM
Call:
   aov(formula = CARMlog ~ Pop + Treat + Pop:Treat, data = dCt)

Terms:
                      Pop     Treat Pop:Treat Residuals
Sum of Squares   0.338406  8.439441  1.039067 23.278895
Deg. of Freedom         2         2         4        63

Residual standard error: 0.6078703
Estimated effects may be unbalanced

> TukeyHSD(CARM)
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = CARMlog ~ Pop + Treat + Pop:Treat, data = dCt)

$Pop
           diff        lwr       upr     p adj
N-H -0.12135852 -0.5425605 0.2998435 0.7692541
S-H -0.16120004 -0.5824021 0.2600020 0.6306273
S-N -0.03984152 -0.4610435 0.3813605 0.9719945

$Treat
         diff        lwr       upr     p adj
M-C 0.1654167 -0.2557853 0.5866187 0.6155473
T-C 0.7947077  0.3735057 1.2159097 0.0000793
T-M 0.6292910  0.2080890 1.0504930 0.0018717

$`Pop:Treat`
                diff         lwr       upr     p adj
N:C-H:C  0.011176998 -0.96506913 0.9874231 1.0000000
S:C-H:C  0.072737132 -0.90350900 1.0489833 0.9999996
H:M-H:C  0.511369332 -0.46487679 1.4876155 0.7546124
N:M-H:C  0.127539081 -0.84870705 1.1037852 0.9999697
S:M-H:C -0.058744224 -1.03499035 0.9175019 0.9999999
H:T-H:C  0.815227728 -0.16101840 1.7914739 0.1753203
N:T-H:C  0.823805424 -0.15244070 1.8000516 0.1651898
S:T-H:C  0.829004028 -0.14724210 1.8052502 0.1592686
S:C-N:C  0.061560134 -0.91468599 1.0378063 0.9999999
H:M-N:C  0.500192334 -0.47605379 1.4764385 0.7759021
N:M-N:C  0.116362083 -0.85988404 1.0926082 0.9999851
S:M-N:C -0.069921222 -1.04616735 0.9063249 0.9999997
H:T-N:C  0.804050729 -0.17219540 1.7802969 0.1892039
N:T-N:C  0.812628426 -0.16361770 1.7888746 0.1784796
S:T-N:C  0.817827030 -0.15841910 1.7940732 0.1722027
H:M-S:C  0.438632200 -0.53761393 1.4148783 0.8764125
N:M-S:C  0.054801949 -0.92144418 1.0310481 1.0000000
S:M-S:C -0.131481356 -1.10772748 0.8447648 0.9999617
H:T-S:C  0.742490595 -0.23375553 1.7187367 0.2798489
N:T-S:C  0.751068292 -0.22517784 1.7273144 0.2657798
S:T-S:C  0.756266896 -0.21997923 1.7325130 0.2574771
N:M-H:M -0.383830251 -1.36007638 0.5924159 0.9382158
S:M-H:M -0.570113556 -1.54635968 0.4061326 0.6323827
H:T-H:M  0.303858395 -0.67238773 1.2801045 0.9845798
N:T-H:M  0.312436092 -0.66381004 1.2886822 0.9816263
S:T-H:M  0.317634695 -0.65861143 1.2938808 0.9796382
S:M-N:M -0.186283305 -1.16252943 0.7899628 0.9994746
H:T-N:M  0.687688647 -0.28855748 1.6639348 0.3799618
N:T-N:M  0.696266343 -0.27997978 1.6725125 0.3632054
S:T-N:M  0.701464947 -0.27478118 1.6777111 0.3532327
H:T-S:M  0.873971951 -0.10227418 1.8502181 0.1146234
N:T-S:M  0.882549648 -0.09369648 1.8587958 0.1073719
S:T-S:M  0.887748252 -0.08849788 1.8639944 0.1031616
N:T-H:T  0.008577696 -0.96766843 0.9848238 1.0000000
S:T-H:T  0.013776300 -0.96246983 0.9900224 1.0000000
S:T-N:T  0.005198604 -0.97104752 0.9814447 1.0000000

kubu4 commented 7 years ago

Just adding the plot of the above Tukey's output to aid in discussion.

plot_tukcarm

kubu4 commented 7 years ago

Use summary(GENE) to view P values of interactions.

kubu4 commented 7 years ago

Does this figure jive with the ANOVA output? My specific concern is what the asterisks indicate (see figure caption below). Does the ANOVA output indicate what is described?

h2av_expression_jh

Figure 3. Expression of H2AV mRNA. Median ΔCt indicated by line in middle of box plot. Shaded boxes are 2nd and 3rd quartile groups. Lines are 1st and 3rd quartiles. Dots indicate outside values. Asterisks indicate significant differences (p<0.05) between treatments within a population. Capital letters indicate significant differences (p<0.05) between overall treatment groups.

ANOVA Output:

> H2AV<-aov(H2AVlog~Pop+Treat+Pop:Treat, data=dCt)
> H2AV
Call:
   aov(formula = H2AVlog ~ Pop + Treat + Pop:Treat, data = dCt)

Terms:
                     Pop    Treat Pop:Treat Residuals
Sum of Squares   2.53478  8.42385   1.69976  38.67588
Deg. of Freedom        2        2         4        63

Residual standard error: 0.7835195
Estimated effects may be unbalanced
> TukeyHSD(H2AV)
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = H2AVlog ~ Pop + Treat + Pop:Treat, data = dCt)

$Pop
           diff        lwr       upr     p adj
N-H -0.36274702 -0.9056589 0.1801649 0.2515685
S-H -0.42578010 -0.9686920 0.1171318 0.1521727
S-N -0.06303307 -0.6059450 0.4798788 0.9581202

$Treat
         diff          lwr       upr     p adj
M-C 0.2706884 -0.272223489 0.8136003 0.4595047
T-C 0.8220292  0.279117331 1.3649411 0.0016085
T-M 0.5513408  0.008428917 1.0942527 0.0457664

$`Pop:Treat`
                diff          lwr       upr     p adj
N:C-H:C -0.289388948 -1.547729628 0.9689517 0.9979850
S:C-H:C -0.603254656 -1.861595336 0.6550860 0.8322637
H:M-H:C  0.261229844 -0.997110837 1.5195705 0.9990296
N:M-H:C -0.003898953 -1.262239633 1.2544417 1.0000000
S:M-H:C -0.337909251 -1.596249931 0.9204314 0.9941297
H:T-H:C  0.727371323 -0.530969358 1.9857120 0.6450102
N:T-H:C  0.193647994 -1.064692687 1.4519887 0.9998943
S:T-H:C  0.652424784 -0.605915897 1.9107655 0.7646261
S:C-N:C -0.313865708 -1.572206388 0.9444750 0.9964502
H:M-N:C  0.550618792 -0.707721889 1.8089595 0.8915600
N:M-N:C  0.285489995 -0.972850685 1.5438307 0.9981691
S:M-N:C -0.048520303 -1.306860983 1.2098204 1.0000000
H:T-N:C  1.016760270 -0.241580410 2.2751010 0.2093831
N:T-N:C  0.483036941 -0.775303739 1.7413776 0.9458977
S:T-N:C  0.941813732 -0.316526949 2.2001544 0.2999651
H:M-S:C  0.864484499 -0.393856181 2.1228252 0.4142086
N:M-S:C  0.599355703 -0.658984978 1.8576964 0.8371478
S:M-S:C  0.265345405 -0.992995275 1.5236861 0.9989139
H:T-S:C  1.330625978  0.072285298 2.5889667 0.0303854
N:T-S:C  0.796902649 -0.461438031 2.0552433 0.5262916
S:T-S:C  1.255679439 -0.002661241 2.5140201 0.0508972
N:M-H:M -0.265128797 -1.523469477 0.9932119 0.9989203
S:M-H:M -0.599139094 -1.857479775 0.6592016 0.8374170
H:T-H:M  0.466141479 -0.792199201 1.7244822 0.9557834
N:T-H:M -0.067581850 -1.325922530 1.1907588 1.0000000
S:T-H:M  0.391194940 -0.867145740 1.6495356 0.9846958
S:M-N:M -0.334010298 -1.592350978 0.9243304 0.9945717
H:T-N:M  0.731270275 -0.527070405 1.9896110 0.6384388
N:T-N:M  0.197546946 -1.060793734 1.4558876 0.9998771
S:T-N:M  0.656323737 -0.602016944 1.9146644 0.7588271
H:T-S:M  1.065280573 -0.193060107 2.3236213 0.1621387
N:T-S:M  0.531557244 -0.726783436 1.7898979 0.9093701
S:T-S:M  0.990334034 -0.268006646 2.2486747 0.2388825
N:T-H:T -0.533723329 -1.792064009 0.7246174 0.9074459
S:T-H:T -0.074946539 -1.333287219 1.1833941 0.9999999
S:T-N:T  0.458776790 -0.799563890 1.7171175 0.9596646

> summary(H2AV)
            Df Sum Sq Mean Sq F value  Pr(>F)   
Pop          2   2.53   1.267   2.064 0.13538   
Treat        2   8.42   4.212   6.861 0.00202 **
Pop:Treat    4   1.70   0.425   0.692 0.60012   
Residuals   63  38.68   0.614                   
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

kubu4 commented 7 years ago

I wasn't entirely sure where to look for this info. @sr320 pointed out that the comparison mentioned in the figure legend would be found in this section of the ANOVA output:

$`Pop:Treat`

For this specific example (indicated by the asterisks), we want this line from the above mentioned section:

S:T-S:C 1.255679439 -0.002661241 2.5140201 0.0508972

This shows the Oyster Bay ("S") population and the comparison between Temperature ("T") and Control ("C"). However, the p-value isn't actually below 0.05...

kubu4 commented 7 years ago

BMP2 and GRB2 both show a Population:Treatment interaction. What do I say about this information?

> summary(BMP2)
            Df Sum Sq Mean Sq F value Pr(>F)  
Pop          2   1.55  0.7747   1.302 0.2793  
Treat        2   1.24  0.6181   1.039 0.3599  
Pop:Treat    4   6.87  1.7176   2.886 0.0293 *
Residuals   63  37.50  0.5952                 
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

> summary(GRB2)
            Df Sum Sq Mean Sq F value Pr(>F)  
Pop          2   2.74  1.3717   2.622 0.0805 .
Treat        2   2.00  1.0012   1.914 0.1559  
Pop:Treat    4   6.92  1.7291   3.306 0.0160 *
Residuals   63  32.95  0.5231                 
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

BMP2 ANOVA shows only the following treatment as being significant:

S:M-S:C -1.28660065 -2.5255863 -0.04761501 0.0359437

GRB2 ANOVA shows the following as being significant:

S:M-N:C -1.28233951 -2.4438505 -0.12082846 0.0198862
S:M-S:C -1.19272883 -2.3542399 -0.03121779 0.0397471
S:M-N:M -1.47959102 -2.6411021 -0.31807998 0.0037183

kubu4 commented 7 years ago

Discussed with Brent. Brief notes:

BMP2: Interaction driven by expression differences between control and mechanical treatment in Oyster Bay pop. Likely due to the cushy environment (i.e. low stress) oysters experience in Oyster Bay.
GRB2: Interaction is complicated, but again, likely due to cushy environment (i.e. low stress) oysters experience in Oyster Bay. Notice that it is the mechanical stress treatment in just the Oyster Bay pop that is contributing to the interactions.

RobertsLab / paper-Olurida-gene

Gene expression analysis ANOVA #11