Introduction
For many years, scientists have understood some of the effects a lack of serotonin has on human behavior. A serotonin deficiency has been credited with numerous unipolar neurological disorders. However, the total effect that a deficiency of serotonin has on human social behaviors is not completely known. The first part of this experiment was aimed at analyzing what those effects are. Using TPEX wild-type mice with normal levels of serotonin and 5HT-Phypo mice with decreased levels of serotonin, I tested the differences in social behaviors between the two genotypes. Preliminary research showed that a lower level of serotonin resulted in increased levels of aggression, which was seen in the heighted number of fights instigated by the 5HT-Phypo mice. To test the validity of these findings I paired unfamiliar wild-type mice with each other, unfamiliar 5HT-Phypo mice with each other, and I also paired wild-type mice with 5HT-Phypo mice. This way I could see how those with normal levels of serotonin and those with diminished levels of serotonin interact with those of their same genotype and those of different genotypes. The experiment used is called a dynamic affiliation test. In a dynamic affiliation test I place each test mouse in clean viewing cage and record the mice on video for three hours and ten minutes. Once the videos are recorded I can score the mice based on a number of social behavior variables. The different variables of importance for this study are allo-sniffing, sleeping together, sleeping alone, and fighting. Allo-sniffing is a control variable as it shows evidence of normal mouse interaction. Sleeping together is a positive affiliation variable. Lastly, fighting is a negative affiliation variable used to observe the level of aggression in the mice. These variables are scored the first ten minutes of every hour and the last ten minutes of the recording. The rest of the 3 hour, 10 minutes recording is scored solely for fight count. This is used to see the overall aggression that accompanies each genotype.
My hypothesis is that the lower level of serotonin will effectively increase the aggressive behavior in the mice. I am looking to see if there is sufficient evidence to support this. At the same time, I have noticed that the 5HT-Phypo mice also show a lack of friendliness to go along with the increased aggression. I will look to the positive affiliation tests to see if there is significant evidence of this effect as well. I hypothesize that lower serotonin lowers positive affiliation.
The second part of my experiment concerns treatment of low serotonin. For years, depression, borderline personality disorder, and many other unipolar disorders have been treated with selective serotonin reuptake inhibitors (SSRIs). The results have been unsatisfactory to say the least. I started looking to see if there was a more effective way to reverse the effects of low serotonin. I reran the experiment having treated the mice with Prozac or Prozac with a 5HT-P booster. In this way I could find whether Prozac alone is effective, or whether a combination of Prozac and 5HT-P is more effective. At the same time, if none of the two treatment methods show signs of effectiveness, this would be evidence that serotonin deficiency cannot be reversed as symptoms are not just the result of genotype, but may be the result of environmental factors that effected the maturation within each specific genotype. I hypothesized that the combination of the two drugs will have the greatest effect on reversing the signs of serotonin deficiency.
Dataset Description
In order for the datasets to be readable in R, I have omitted the column headings. Instead, I used vectors with numbers. Vector 1 corresponds to mouse pair number. Vector 2 corresponds to mouse 1 genotype. Vector 3 corresponds to mouse 2 genotype. Vector 4 corresponds to treatment stage whether pre or post-treatment. Vector 5 corresponds to fight count. Vector 6 corresponds to count of sleeping together occurrence. Mouse1 dataset compares the fight and sleeping together occurrence counts comparing the wild-type and 5HT-p(Ho) pre treatment. Mouse2 dataset compares the fight and sleeping together occurrence counts comparing the 5HT-p(Ho) mice pre and post treatment. Mouse3 dataset compares the fight and sleeping together occurrence counts comparing the wild-type/5HT-p(Ho) pairs pre and post treatment. I have included boxplots below for the sae of visualizing the results of these comparisons.
mouse1 <- read.csv("~/mouse1.csv", header = F)
head(mouse1, n = 18)
## V1 V2 V3 V4 V5 V6
## 1 1 Ho Ho pre 10 0
## 2 2 Ho Ho pre 52 0
## 3 3 Ho Ho pre 16 0
## 4 4 Ho Ho pre 19 0
## 5 5 Ho Ho pre 39 0
## 6 6 Ho Ho pre 32 0
## 7 7 Ho Ho pre 78 0
## 8 8 Ho Ho pre 5 0
## 9 9 Ho Ho pre 18 0
## 10 10 Wt Wt pre 0 1
## 11 11 Wt Wt pre 0 6
## 12 12 Wt Wt pre 0 3
## 13 13 Wt Wt pre 0 0
## 14 14 Wt Wt pre 0 0
## 15 15 Wt Wt pre 0 0
## 16 16 Wt Wt pre 14 0
## 17 17 Wt Wt pre 0 1
## 18 18 Wt Wt pre 0 0
mouse2 <- read.csv("~/mouse2.csv", header = F)
head(mouse2, n = 18)
## V1 V2 V3 V4 V5 V6
## 1 1 Ho Ho pre 10 0
## 2 2 Ho Ho pre 52 0
## 3 3 Ho Ho pre 16 0
## 4 4 Ho Ho pre 19 0
## 5 5 Ho Ho pre 39 0
## 6 6 Ho Ho pre 32 0
## 7 7 Ho Ho pre 78 0
## 8 8 Ho Ho pre 5 0
## 9 9 Ho Ho pre 18 0
## 10 10 Ho Ho post 0 0
## 11 11 Ho Ho post 4 1
## 12 12 Ho Ho post 0 0
## 13 13 Ho Ho post 0 0
## 14 14 Ho Ho post 0 0
## 15 15 Ho Ho post 0 2
## 16 16 Ho Ho post 2 4
## 17 17 Ho Ho post 0 1
## 18 18 Ho Ho post 0 5
mouse3 <- read.csv("~/mouse3.csv", header = F)
head(mouse2, n = 14)
## V1 V2 V3 V4 V5 V6
## 1 1 Ho Ho pre 10 0
## 2 2 Ho Ho pre 52 0
## 3 3 Ho Ho pre 16 0
## 4 4 Ho Ho pre 19 0
## 5 5 Ho Ho pre 39 0
## 6 6 Ho Ho pre 32 0
## 7 7 Ho Ho pre 78 0
## 8 8 Ho Ho pre 5 0
## 9 9 Ho Ho pre 18 0
## 10 10 Ho Ho post 0 0
## 11 11 Ho Ho post 4 1
## 12 12 Ho Ho post 0 0
## 13 13 Ho Ho post 0 0
## 14 14 Ho Ho post 0 0
boxplot(mouse1$V5 ~ mouse1$V2, main = "Fight Count Comparison Pre-Treatment",
ylab = "Fight Count", xlab = "Genotype")
boxplot(mouse2$V5 ~ mouse2$V4, main = "Fight Count Comparison Post/Pre-Treatment 5HT-P Mice",
ylab = "Fight Count", xlab = "Treatment Stage")
boxplot(mouse3$V5 ~ mouse3$V4, main = "Fight Count Comparison Post/Pre-Treatment Wt/5HT-P Mice",
ylab = "Fight Count", xlab = "Treatment Stage")
boxplot(mouse1$V6 ~ mouse1$V2, main = "Sleep Together Count Comparison Pre-Treatment",
ylab = "Sleeping Together Occurrence Count", xlab = "Genotype")
boxplot(mouse2$V6 ~ mouse2$V4, main = "Sleep Together Comparison Post/Pre-Treatment 5HT-P Mice",
ylab = "Sleeping Together Occurrence Count", xlab = "Treatment Stage")
boxplot(mouse3$V6 ~ mouse3$V4, main = "Sleep Together Comparison Post/Pre-Treatment Wt/5HT-P Mice",
ylab = "Sleeping Together Occurrence Count", xlab = "Treatment Stage")
Data Analysis
In order to complete hypothesis testing, I reorganized the data, comparing the mean fight counts and mean sleeping together occurrence counts for my three comparisons. When analyzing the difference between the two genotypes before treatment the null hypothesis, is that there is no difference between the mean fight counts and mean sleeping together occurrence counts for the two genotypes. The alternative hypothesis is that there is a difference between the mean fight counts and mean sleeping together occurrence counts for the two genotypes, in that the 5HT-P mice have on average more fights than the Wt mice. When analyzing the difference between the 5HT-P mice before and after treatment the null hypothesis, is that there is no difference between the mean fight counts and mean sleeping together occurrence counts for the two treatment stages. The alternative hypothesis is that there is a difference between the mean fight counts and mean sleeping together occurrence counts for the two treatment stages, in that the treated 5HT-P mice have on average fewer fights than the untreated 5HT-P mice. When analyzing the difference between the Wt/5HT-P mouse pairs before and after treatment the null hypothesis, is that there is no difference between the mean fight counts and mean sleeping together occurrence counts for the two treatment stages. The alternative hypothesis is that there is a difference between the mean fight counts and mean sleeping together occurrence counts for the two treatment stages, in that the treated mouse pairs have on average fewer fights than the untreated mouse pairs. All conditions for hypothesis testing are assumed to apply to the datasets in question. A significance level of .05 is used in all analyses. For each comparison, I will use a t-test to test for sufficient evidence to support the alternative hypothesese.
# Here is the data comparing the Wt mice to the 5HT-P mice before
# treatment
means1 <- read.delim("~/MeansHoVsWtPreCompar.csv")
head(means1, n = 4)
## X Ho.pre.treatment.fight.count Wt.pre.treatment.fight.count
## 1 n 9.00 9.00
## 2 x? 29.88 1.56
## 3 s 22.00 4.39
## 4 SE 2.44 0.48
## Ho.pre.treatment.Sleeping.Together.Occurrence.Count
## 1 9
## 2 0
## 3 0
## 4 0
## Wt.pre.treatment.Sleeping.Together.Occurrence.Count
## 1 9.00
## 2 1.22
## 3 1.93
## 4 0.21
# T-test statistic for fight counts
((29.88 - 1.56) - 0)/7.48
## [1] 3.786
# T-test statistic for sleeping together occurrence counts
((0 - 1.22) - 0)/0.64
## [1] -1.906
Comparing the two genotypes before treatment, the test statistic for the fight count comparison is 3.786096 with 8 degrees of freedom. This yields a p-value of 0.00534153. Due to the fact that this p-value is far below my significance level, I conclude that there is significant evidence to support the notion that the 5HT-P mice fight more on average than the Wt mice. Comparing the two genotypes before treatment, the test statistic for the sleeping together occurrence count comparison is -1.90625 with 8 degrees of freedom. This yields a p-value of 0.09307465. Due to the fact that this p-value is not less than my significance level, I cannot conclude that genotype has any bearing on the average number of sleeping together occurrences before treatment.
# Here is the data comparing the 5HT-P mice before and after treatment
means2 <- read.delim("~/MeansHoHoPrePostCompar.csv")
head(means2, n = 4)
## X Ho.pre.treatment.fight.count Ho.post.treatment.fight.count
## 1 n 9.00 9.00
## 2 x? 29.88 0.66
## 3 s 22.00 1.33
## 4 SE 2.44 0.15
## Ho.pre.treatment.Sleeping.Together.Occurrence.Count
## 1 9
## 2 0
## 3 0
## 4 0
## Ho.post.treatment.Sleeping.Together.Occurrence.Count
## 1 9.00
## 2 1.44
## 3 1.77
## 4 0.20
# T-test statistic for fight counts
((29.88 - 0.66) - 0)/7.35
## [1] 3.976
# T-test statistic for sleeping together occurrence counts
((0 - 1.44) - 0)/0.59
## [1] -2.441
Comparing the 5HT-P mice before and after treatment, the test statistic for the fight count comparison is 3.97551 with 8 degrees of freedom. This yields a p-value of 0.00408729. Due to the fact that this p-value is far below my significance level, I conclude that there is significant evidence to support the notion that the 5HT-P mice fight more on average before treatment than after treatment. Comparing the the 5HT-P mice before and after treatment, the test statistic for the sleeping together occurrence count comparison is -2.440678 with 8 degrees of freedom. This yields a p-value of 0.04052656. Due to the fact that this p-value is less than my significance level, I can conclude that treatment increases the average number of sleeping together occurrences in 5HT-P mice.
# Here is the data comparing the Wt/5HT-P mouse pairs before and after
# treatment
means3 <- read.delim("~/MeansHoWtPrePostCompar.csv")
head(means3, n = 4)
## X X.Ho.Wt.pre.treatment.fight.count X.Ho.Wt.post.treatment.fight.count
## 1 n 7.00 7.00
## 2 x? 9.71 4.29
## 3 s 6.32 10.49
## 4 SE 2.39 3.97
## X.Ho.Wt.pre.treatment.Sleeping.Together.Occurrence.Count
## 1 7
## 2 0
## 3 0
## 4 0
## X.Ho.Wt.post.treatment.Sleeping.Together.Occurrence.Count
## 1 7.00
## 2 2.57
## 3 2.49
## 4 1.58
# T-test statistic for fight counts
((9.71 - 4.29) - 0)/4.63
## [1] 1.171
# T-test statistic for sleeping together occurrence counts
((0 - 2.57) - 0)/0.94
## [1] -2.734
Comparing the Wt/5HT-P mouse pairs before and after treatment, the test statistic for the fight count comparison is 1.170626 with 7 degrees of freedom. This yields a p-value of 0.28005604. Due to the fact that this p-value is not below my significance level, I cannot conclude that there is significant evidence to support the notion that the Wt/5HT-P mouse pairs fight more on average before treatment than they do after treatment. Comparing the the Wt/5HT-P mouse pairs before and after treatment, the test statistic for the sleeping together occurrence count comparison is -2.734043 with 7 degrees of freedom. This yields a p-value of 0.02916972. Due to the fact that this p-value is less than my significance level, I can conclude that treatment increases the average number of sleeping together occurrences in Wt/5HT-P mouse pairs.