By: Connor Brown
The purpose of this project was to compare levels of gene expression between cocaine addicts and a control group. The genes with at least 2-fold increase in expression in the cocaine group were identified and classified by function.
Above is the top left corner of our dataframe. The rows represent patients, and the columns are codes for mRNA transcripts. The numbers underneath the genes represent relative levels of that transcript, and can be viewed as # of transcripts. The mRNA was extracted from midbrain dopamine (DA)-synthesizing neurons of 30 deceased cocaine addicts, and 30 deceased control subjects.
Above is a histogram and boxplot of the expression levels for all genes including cocaine and control group. The IQR for all expression levels is between 135.7 to 240.8, the median expression level is 145.5, and above 400 is considered to be outliers. There are 8542 outliers. There is strong right skewness as represented by the positive g1 value above.
Above is a histogram and boxplot for the absolute value of the expression level differences between the cocaine group and the control group. The IQR is between 1.034% and 3.876% for the mRNA's, and the mean difference is 2.248%. Anything above 10% change is considered to be outliers. There are 4345 outliers. There is strong, right skewness as indicated by the positive value for g1.
Above is the left side of a table containing the medians, means, skewness, and kurtosis for both the cocaine group and control group for all of the mRNA's. You can search by your mRNA of choice easily in R, as shown by typing:
B[1:11, "NRP2"]
Just replace the "NRP2" with your mRNA of choice.
Above are the 28 mRNA's identified that have at least 2-fold expression in the cocaine group than the control group.
I then went online and did research on the mRNA's and noticed that many of their functions had similar roles. I classified them by categories(Cell Cycle Progression/Differentation, Immune/Inflammatory Response, Protein Degradation, and Misc). Above is an image of an excel spreadsheet with the classifications, and brief descriptions of function.
Overall, 10 of the mRNA's fell into Cell Cycle Progression/Differentation, 8 into Immune/Inflammatory Response, 6 into Protein Degradation, and 4 into Misc.
Because FOS, FOSB, and JUN (Cell Cycle Progression/Differentation) all form the AP-1 transcription factor complex together, I made confidence intervals for the three of them (cocaine and control). The results can be seen above.
Next, I tested for statistical significance between the expression levels for the 3 mRNA's between the cocaine and control groups. I used alpha = 0.05.
Above is the 2-mean hypothesis test for cocaine$FOS and control$FOS. Because the test statistic is greater than tα/2,29, we reject the null hypothesis that the two means are equal.
Above is the 2-mean hypothesis test for cocaine$FOSB and control$FOSB. Because the test statistic is greater than tα/2,29, we reject the null hypothesis that the two means are equal.
Above is the 2-mean hypothesis test for cocaine$JUN and control$JUN. Because the test statistic is greater than tα/2,29, we reject the null hypothesis that the two means are equal.
To compute ANOVA, I re-named the 28 mRNA's to their classifications to test for statistical significance between the mean expression differences. Reminder: A difMeans value of 1.203 means that there is a 120.3% increase in expression for that mRNA between the cocaine group and the control group.
The null hypothesis is that the mean expression change is the same for all classifications. The alternate hypothesis is that the means are not the same. The p-value is 0.422, so we fail to reject the null hypothesis.
I wanted to see if the difference between protein and immune was statistically significant, but after conducting a t test that returned a p-value of 0.4131, we fail to reject the null hypothesis.
THANK YOU FOR READING!