Showing posts with label statistics. Show all posts
Showing posts with label statistics. Show all posts

Wednesday, July 22, 2015

7 22 2015 BMP2 qPCR 2

Today I ran duplicates for the targets I ran last week. I'm hoping to get strong replicates for analysis. Here I ran the BMP2 which last week had issues with the NTC. 

Primers:

1640BMP-2_FWDTGAAGGAACGACCAAAGCCAJH5/21/20152055O.luridaBone morphogenetic protein 2 (BMP-2) (Bone morphogenetic protein 2A) (BMP-2A)P12643
1639BMP-2_REVTCCGGTTGAAGAACCTCGTGJH5/21/20152055O.luridaBone morphogenetic protein 2 (BMP-2) (Bone morphogenetic protein 2A) (BMP-2A)P12643


Reagent Table:
VolumeReactions X116
Ssofast Evagreen MM101160
FWD Primer0.558
REV Primer0.558
1:9 cDNA9
  1. Added reagents from greatest to least volume
  2. Vortexed
  3. Centrifuged briefly
  4. Pipetted 11 ul Master Mix to each tube
  5. Pipetted 9 ul of 1:9 cDNA each column using a channel pipetter
  6. Centrifuged plate at 2000 rpm for 1 minute
  7. Ran Program Below
Program:
StepTemperatureTime
Initiation95 C10 min
Elongation95 C30 sec
60 C1 min
Read
72 C30 sec
Read
Repeat Elongation 39 times
Termination95 C1 min
55 C1 sec
Melt Curve Manual ramp 0.2C per sec Read 0.5 C55 - 95 C30 sec
21 C10 min
End
Plate Layout:
1234567
DNased 42215 HC1DNased 42215 NC1DNased 42215 SC1DNased 42215 HT1 1DNased 42215 NT1 1DNased 42215 ST1 1NTC
DNased 42215 HC2DNased 42215 NC2DNased 42215 SC2DNased 42215 HT1 2DNased 42215 NT1 2DNased 42215 ST1 2NTC
DNased 42215 HC3DNased 42215 NC3DNased 42215 SC3DNased 42215 HT1 3DNased 42215 NT1 3DNased 42215 ST1 3NTC
DNased 42215 HC4DNased 42215 NC4DNased 42215 SC4DNased 42215 HT1 4DNased 42215 NT1 4DNased 42215 ST1 4NTC
DNased 42215 HC5DNased 42215 NC5DNased 42215 SC5DNased 42215 HT1 5DNased 42215 NT1 5DNased 42215 ST1 5
DNased 42215 HC6DNased 42215 NC6DNased 42215 SC6DNased 42215 HT1 6DNased 42215 NT1 6DNased 42215 ST1 6
DNased 42215 HC7DNased 42215 NC7DNased 42215 SC7DNased 42215 HT1 7DNased 42215 NT1 7DNased 42215 ST1 7
DNased 42215 HC8DNased 42215 NC8DNased 42215 SC8DNased 42215 HT1 8DNased 42215 NT1 8DNased 42215 ST1 8
Results:

All samples
NTCs

These amplification curves look better than the last run of BMP2. There is some minor amplification in the NTCs but it is larger than the target and is probably an artifact of the PCR process. To analyze these data I ran them through an updated stats and graphs script which runs a Two Way ANOVA and produces a boxplot and expression bar graph. The data did not have its outliers eliminated which worked fine with this as most samples showed some expression. 

Expression Bar Graph


Statistics

Call:
   aov(formula = expression ~ Pop + Treat + Pop:Treat, data = rep2res2)

Terms:
                         Pop        Treat    Pop:Treat    Residuals
Sum of Squares  1.703293e-20 3.032145e-20 2.201442e-20 6.077281e-20
Deg. of Freedom            2            1            2           42

Residual standard error: 3.803908e-11
Estimated effects may be unbalanced
> TukeyHSD(fit)
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = expression ~ Pop + Treat + Pop:Treat, data = rep2res2)

$Pop
            diff           lwr          upr     p adj
N-H  0.4089519
S-H  0.0041846
S-N  0.1002050

$Treat
             diff           lwr           upr    p adj
T-C  4.14e-05

$`Pop:Treat`
                 diff           lwr           upr     p adj
N:C-H:C   0.6348257
S:C-H:C   0.0001275
H:T-H:C  0.9973637
N:T-H:C  0.9999701
S:T-H:C  0.9814628
S:C-N:C   0.0134535
H:T-N:C  0.3564950
N:T-N:C  0.5192134
S:T-N:C  0.2393859
H:T-S:C  0.0000292
N:T-S:C  0.0000710
S:T-S:C  0.0000133
N:T-H:T   0.9997768
S:T-H:T  0.9998827
S:T-N:T  0.9953619

There definitely appears a significant difference between treatment and control. From the stats it appears the Oyster Bay population control has significantly larger expression of BMP2 than the other two populations as well as it the Oyster Bay treatment. You can see this in the bar graph generated below.


The most distressing thing about these data is that it doesn't match the graphs from last week. The controls have much lower expression than they did last week which should not have changed between replicates. I reran the previous data with all the outliers but it still doesn't change the statistics. 

Previous Expression Data

Previous Data Stats
Call:
   aov(formula = expression ~ Pop + Treat + Pop:Treat, data = rep2res2)

Terms:
                         Pop        Treat    Pop:Treat    Residuals
Sum of Squares  3.918260e-22 9.723005e-21 1.573763e-21 9.279358e-21
Deg. of Freedom            2            1            2           38

Residual standard error: 1.56267e-11
Estimated effects may be unbalanced
4 observations deleted due to missingness
> TukeyHSD(fit)
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = expression ~ Pop + Treat + Pop:Treat, data = rep2res2)

$Pop
                 p adj
N-H   0.4224591
S-H   0.8174685
S-N  0.8189700

$Treat
         p adj
T-C  2e-07

$`Pop:Treat`
                    p adj
N:C-H:C   0.6961640
S:C-H:C  0.9298189
H:T-H:C  0.0032438
N:T-H:C  0.0048368
S:T-H:C  0.2481527
S:C-N:C  0.1806352
H:T-N:C  0.0000405
N:T-N:C  0.0000732
S:T-N:C  0.0140727
H:T-S:C  0.0411016
N:T-S:C  0.0532668
S:T-S:C  0.7297433
N:T-H:T   1.0000000
S:T-H:T   0.7788719
S:T-N:T   0.7991437

The stats don't match which is distressing. Even the boxplots dont seem to match. 

You can see the raw data files for this week here and last week here

Monday, July 20, 2015

7 20 2015 p29ING qPCR

Today I ran more targets that may be of interest due expected changes in expression. Here I explore p29ING which is expected to have increased expression due to exposure to increased temperatures. 

Primers:


1624p29ING4_FWDTACCTTTGGGCTTCACCGTCJH5/21/20152055O.luridaInhibitor of growth protein 4 (p29ING4)Q8C0D7
1623p29ING4_REVGTCCATCACACACCCCTCAGJH5/21/20152055O.luridaInhibitor of growth protein 4 (p29ING4)Q8C0D7

Reagent Table:
VolumeReactions X116
Ssofast Evagreen MM101160
FWD Primer0.558
REV Primer0.558
1:9 cDNA9
  1. Added reagents from greatest to least volume
  2. Vortexed
  3. Centrifuged briefly
  4. Pipetted 11 ul Master Mix to each tube
  5. Pipetted 9 ul of 1:9 cDNA each column using a channel pipetter
  6. Centrifuged plate at 2000 rpm for 1 minute
  7. Ran Program Below
Program:
StepTemperatureTime
Initiation95 C10 min
Elongation95 C30 sec
60 C1 min
Read
72 C30 sec
Read
Repeat Elongation 39 times
Termination95 C1 min
55 C1 sec
Melt Curve Manual ramp 0.2C per sec Read 0.5 C55 - 95 C30 sec
21 C10 min
End
Plate Layout:
1234567
DNased 42215 HC1DNased 42215 NC1DNased 42215 SC1DNased 42215 HT1 1DNased 42215 NT1 1DNased 42215 ST1 1NTC
DNased 42215 HC2DNased 42215 NC2DNased 42215 SC2DNased 42215 HT1 2DNased 42215 NT1 2DNased 42215 ST1 2NTC
DNased 42215 HC3DNased 42215 NC3DNased 42215 SC3DNased 42215 HT1 3DNased 42215 NT1 3DNased 42215 ST1 3NTC
DNased 42215 HC4DNased 42215 NC4DNased 42215 SC4DNased 42215 HT1 4DNased 42215 NT1 4DNased 42215 ST1 4NTC
DNased 42215 HC5DNased 42215 NC5DNased 42215 SC5DNased 42215 HT1 5DNased 42215 NT1 5DNased 42215 ST1 5
DNased 42215 HC6DNased 42215 NC6DNased 42215 SC6DNased 42215 HT1 6DNased 42215 NT1 6DNased 42215 ST1 6
DNased 42215 HC7DNased 42215 NC7DNased 42215 SC7DNased 42215 HT1 7DNased 42215 NT1 7DNased 42215 ST1 7
DNased 42215 HC8DNased 42215 NC8DNased 42215 SC8DNased 42215 HT1 8DNased 42215 NT1 8DNased 42215 ST1 8
Results:

All samples
NTCs
These amplification curves look much better than the GRB2. There is some minor amplification in 2 of the 4 NTCs but the product was smaller than the target. The one weird issue is that one sample has a double peak in the melt curve. I'm not sure what caused this as it doesn't appear anywhere else. It may be something odd with the gene being expressed. I then analyzed the data with my stats and graphs script to see what the significant differences are. 

Adjusted Expression Bargraph


Adjusted Statistics

One Way ANOVA comparing Population Controls 

Call:
   aov(formula = expression ~ Pop, data = rep2res2[Treat == "C"])

Terms:
                         Pop    Residuals
Sum of Squares  1.026249e-21 2.641735e-21
Deg. of Freedom            2           21

Residual standard error: 1.121592e-11
Estimated effects may be unbalanced
> TukeyHSD(fit2)
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = expression ~ Pop, data = rep2res2[Treat == "C"])

$Pop
             diff           lwr          upr     p adj
N-H  0.1997946
S-H   0.5595058
S-N   0.0263669

Only the Oyster Bay was significantly different from Fidalgo in the control group.The bloxplot represents this very well.  

Adjusted Expression Boxplot

You can see the raw data here

7 20 2015 GRB2 qPCR

Today I ran more targets that may be of interest due expected changes in expression. Here I explore GRB2 which is expected to have decreased expression due to exposure to increased temperatures. 

Primers:

1612GRB2_FWDAACTTTGTCCACCCAGACGGJH5/21/20152055O.luridaGrowth factor receptor-bound protein 2 (Adapter protein GRB2) (Protein Ash) (SH2/SH3 adapter GRB2)P62994
1611GRB2_REVCCAGTTGCAGTCCACTTCCTJH5/21/20152055O.luridaGrowth factor receptor-bound protein 2 (Adapter protein GRB2) (Protein Ash) (SH2/SH3 adapter GRB2)P62994


Reagent Table:
VolumeReactions X116
Ssofast Evagreen MM101160
FWD Primer0.558
REV Primer0.558
1:9 cDNA9
  1. Added reagents from greatest to least volume
  2. Vortexed
  3. Centrifuged briefly
  4. Pipetted 11 ul Master Mix to each tube
  5. Pipetted 9 ul of 1:9 cDNA each column using a channel pipetter
  6. Centrifuged plate at 2000 rpm for 1 minute
  7. Ran Program Below
Program:
StepTemperatureTime
Initiation95 C10 min
Elongation95 C30 sec
60 C1 min
Read
72 C30 sec
Read
Repeat Elongation 39 times
Termination95 C1 min
55 C1 sec
Melt Curve Manual ramp 0.2C per sec Read 0.5 C55 - 95 C30 sec
21 C10 min
End
Plate Layout:
1234567
DNased 42215 HC1DNased 42215 NC1DNased 42215 SC1DNased 42215 HT1 1DNased 42215 NT1 1DNased 42215 ST1 1NTC
DNased 42215 HC2DNased 42215 NC2DNased 42215 SC2DNased 42215 HT1 2DNased 42215 NT1 2DNased 42215 ST1 2NTC
DNased 42215 HC3DNased 42215 NC3DNased 42215 SC3DNased 42215 HT1 3DNased 42215 NT1 3DNased 42215 ST1 3NTC
DNased 42215 HC4DNased 42215 NC4DNased 42215 SC4DNased 42215 HT1 4DNased 42215 NT1 4DNased 42215 ST1 4NTC
DNased 42215 HC5DNased 42215 NC5DNased 42215 SC5DNased 42215 HT1 5DNased 42215 NT1 5DNased 42215 ST1 5
DNased 42215 HC6DNased 42215 NC6DNased 42215 SC6DNased 42215 HT1 6DNased 42215 NT1 6DNased 42215 ST1 6
DNased 42215 HC7DNased 42215 NC7DNased 42215 SC7DNased 42215 HT1 7DNased 42215 NT1 7DNased 42215 ST1 7
DNased 42215 HC8DNased 42215 NC8DNased 42215 SC8DNased 42215 HT1 8DNased 42215 NT1 8DNased 42215 ST1 8
Results:

All samples

NTCs

These curves don't look great. The 3/4 NTCs have amplification with a product that is the same size as the target. This qPCR needs to be re run to see if the data is valid. I went ahead and ran the stats script on the data anyway assuming that the NTCs were accidentally contaminated. 

Adjusted Expression Bargraph


Adjusted Statistics Summary

Two Way ANOVA 
Call:
   aov(formula = expression ~ Pop + Treat + Pop:Treat, data = rep2res2)

Terms:
                         Pop        Treat    Pop:Treat    Residuals
Sum of Squares  1.209760e-20 4.221308e-19 7.201400e-21 6.405789e-19
Deg. of Freedom            2            1            2           42

Residual standard error: 1.234985e-10
Estimated effects may be unbalanced

  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = expression ~ Pop + Treat + Pop:Treat, data = rep2res2)

$Pop
                 p adj
N-H   0.9388267
S-H  0.8500743
S-N  0.6540849

$Treat
              p adj
T-C  4.5e-06

$`Pop:Treat`
                 diff           lwr           upr     p adj
N:C-H:C  0.9998943
S:C-H:C  0.9793619
H:T-H:C  0.0109104
N:T-H:C  0.0674973
S:T-H:C  0.0093246
S:C-N:C 0.9967000
H:T-N:C  0.0205525
N:T-N:C  0.1139468
S:T-N:C  0.0176867
H:T-S:C  0.0672949
N:T-S:C  0.2845566
S:T-S:C  0.0589259
N:T-H:T   0.9791965
S:T-H:T  0.9999999
S:T-N:T  0.9709791

One Way ANOVA comparing Population Controls
Call:
   aov(formula = expression ~ Pop, data = rep2res2[Treat == "C"])

Terms:
                         Pop    Residuals
Sum of Squares  8.045000e-21 6.051431e-19
Deg. of Freedom            2           21

Residual standard error: 1.697538e-10
Estimated effects may be unbalanced
> TukeyHSD(fit2)
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = expression ~ Pop, data = rep2res2[Treat == "C"])

$Pop
             diff           lwr          upr     p adj
N-H  0.9840303
S-H  0.8632357
S-N  0.9358047

One Way ANOVA comparing Population Treatments
Call:
   aov(formula = expression ~ Pop, data = rep2res2[Treat == "T"])

Terms:
                         Pop    Residuals
Sum of Squares  1.125392e-20 3.543572e-20
Deg. of Freedom            2           21

Residual standard error: 4.107816e-11
Estimated effects may be unbalanced
> TukeyHSD(fit3)
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = expression ~ Pop, data = rep2res2[Treat == "T"])

$Pop
               p adj
N-H   0.104549
S-H  0.984109
S-N  0.075302

T-Test for DABOB

Welch Two Sample t-test

data:  expression by Treat
t = 2.9537, df = 7.1, p-value = 0.02095
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 4.449457e-11 3.966833e-10
sample estimates:
mean in group C mean in group T 
   2.429139e-10    2.232492e-11 

T-Test for FIDALGO

Welch Two Sample t-test

data:  expression by Treat
t = 2.6905, df = 9.489, p-value = 0.02369
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 2.685431e-11 2.971352e-10
sample estimates:
mean in group C mean in group T 
   2.284042e-10    6.640939e-11 

T-Test for Oyster Bay

Welch Two Sample t-test

data:  expression by Treat
t = 3.8083, df = 7.192, p-value = 0.006313
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 6.887209e-11 2.913015e-10
sample estimates:
mean in group C mean in group T 
   1.989093e-10    1.882251e-11 

There is a strong difference between the treatment and control. This is apparent in every population. The populations though are not statistically different from one another. You can see this in the boxplot generated from the data. 

You can see the raw data file here.