Statics of gene expression data

Question 1

What are the properties of a normal curve

Answer 1

• Symmetry
• Normal distribution
• Bell shaped!
• Centered at u
• Almost all values are included between u-3o to u+3o

u (or x)= average
o=standard deviation

Question 2

What does AUC mean

Answer 2

Area under curve, one of the most important uses of the curve of a variable is that percentages for the variable are EQUAL to the AUC

Question 3

Is there a solution for when you are not able to calculate the AUC with a normal curve?

Answer 3

You can make a STANDARD normal curve

Question 4

How do you go from a normal curve to a standard normal curve

Answer 4

We normalize the curves of the normal distribution to the standard normal curve. X becomes Z because the x value is changed, so you calculate the Z-score.
Z score: substract the u from the x value and then divide it by o for each point of the curve

A normal distribution with a mean of 0 and a standard deviation of 1 is called a standard normal distribution

Question 5

What is the dF T curves?

Answer 5

dF= Degrees of Freedom. A particular T-distribution is identified by its dF

1 sample T-test dF=n-1
2 sample T-test dF= n1+n2-2

Question 6

What are the different types of T-tests and when do you use them?

Answer 6

One sample T-test: Is there a difference between a group and the population?

Unpaired samples t-test: Is there a difference between 2 unrelated groups?

Paired samples t-test: Is there a difference in a group between 2 points in time

Question 7

What is the difference between a 1 tailed t-test and a 2 tailed t-test?

Answer 7

It depends where you’re interested in

1 tailed: Is the average salary of male teachers HIGHER than women? You are focussinf on the answer higher, and ignore possibilities. A 1 tailed t-test is only justified if you habe a specific prediction about the direction of the diffeence

2 tailed: Is the average salary of male teachers DIFFERENT than women?

Question 8

What is the P value

Answer 8

The P value ranges from 0-1 and it represents the probability of obtaining the extreme results than the actual results when the nul hypothesis is true. The lower the p value, it is very unlikely to obtain such extreme results when there is no statistically significant difference between the mean of 2 groups (or the mean of 1 group and reference value)

Question 9

When is something false positive, and when is something false negative and name the type of errors

Answer 9

False positive is Type I error. This is when your result should be negative, but the outcome is positive. In other words; When the null hypothesis is true, but according to the results, it has to be rejected

False negative is Type II error. This is when your result should be positive, but the outcome is negative. In other words; when the null hypothesis is false, but according to the results, it isn’t false so the null hypothesis is not rejected.

Question 10

Which statistic method can help with decreasing the probability on type I errors and explain how it’s done

Answer 10

To have less false positive outcomes, use the Bonferroni correction. Sets the P value at for example 0,0025 instead of 0,05. This will mean that the type II error probability will increase

Question 11

Instead of using Bonferroni correction, FDR can also be used to correct type I errors. How?

Answer 11

FDR= False discovery rate
= The expected proportion of type I errors. Controlling for FDR is a way to identify as many significant features as possible, while incurring a relatively low proportion of type I errors. An FDR of 5% means that, among all features called significant, 5% of these are truly null. P value is called the Q value

Question 12

Are these experiments a paired or unpaired t-test
1. The before and after effects of a treatment on the same group of people

2. Research, such as drug study or other treatment plan, where the half of the subjects are assigned to the treatment group and the other half of the subjects are randomly assigned to the control group
3. Body temperature measuring using 2 different thermometers on the same group
4. Comparing the average commuting distance traveled by NYC and San Fransisco residents using 1000 randomly selected participants from each city
5. Standardizing test results of a group of students before and after a study prep course

Answer 12

1. Paired t-test
2. Unpaired t-test
3. Paired t-test
4. Unpaired t-test
5. Paired t-test