BIO205 week 6 Case study

From Sickle cell genetics to therapy

What will we learn

  • Comparison between means with one-sample t-tests
  • Comparison between means with two-sample t-tests
  • Current Treatment options for treating Sickle Cell Anemia, and the process of scientific inquiry
  • Explore the ethics of gene therapies
  • Collect data for our own, in-class experiment (Thursday)

What is due at the end of this week

  • No homework this week
  • Read grant overview for lab, and set up individual meetings.

From Sickle cell genetics to therapy

Part 1. Fetal hemoglobin might be the answer.

Sickle Cell Disease (SCD).

Last week, we were introduced to the sickle variant of hemoglobin (HbS), which is a mutation in the beta globin gene, that causes SCD. The sickle variant causes the beta globin to form inflexible chains of globins inside red blood cells. Beta globin dimerizes and binds another globin dimer, which produces a 4 globin structure that can carry oxygen throughout the body. Thus, the sickle variant can lead to poor delivery of oxygen to tissues throughout the body, in addition to blood clots.

Figure: Top, Red blood cells contain hemoglobin, which is made of 4 globin polypeptides. In normal adults, two are made of beta globin, and two are made of alpha globin (refered to as HbA). Bottom, Prior to birth (prenatally), hemoglobin is not made of beta globins but rather gamma globins (refered to as HbF). This switch occurs because babies in utero need to bind oxygen with higher affinity, to get oxygen from mom.

Treatments.

Current treatments for SCD include drug treatments and donor bone marrow replacements, with varying success. Newer treatments involve gene therapy. In part 1, we will analyze one of the treatments. In this treatment, scientist target a regulatory gene called BCL11A. This gene acts on the promoter region of the  globin gene, to reduce  globin gene expression and protein abundance. Thus, if they were to inhibit BCL11A, maybe they can increase the amount of fetal hemoglobin (HbF).

The data

The data was collected and is shown below. Note that the normal levels of the measurements are the following: Hemoglobin in g/dl: 11 Hb.g.dl Hematocrit: 35% Percent HbF (HbF/total hemoglobin): 0.6 Percent F cells (RBCs containing HbF/total RBCs): 2.5

# read the file; this is preloaded into your working directory
sickle <- read.csv("SickleCell/BIO205sickle1.csv", header = T)
sickle
##   Patient MonthsSinceInfusion Hb.g.dl Hematocrit PercentHbF
## 1       1                  24    11.4       32.5       22.7
## 2       2                  18     9.5       28.9       20.4
## 3       3                  21    11.1       31.2       31.9
## 4       4                  15    11.0       32.5       38.8
## 5       5                  12    11.0       31.4       29.0
## 6       6                   6     9.3       25.5       41.3
##   PercentFcells_recent PercentFcells_baseline
## 1                 71.0                      1
## 2                 58.9                      9
## 3                 81.9                     19
## 4                 65.3                     23
## 5                 70.6                     26
## 6                 93.6                      7

First, we want to compare the levels of hematocrit (fraction of the blood that is red blood cells) from the 6 patients receiving the transfusion, with expected values of normal, control patients.

We should come up with a statistical null hypothesis for this comparison. Write one down

Now, we have to figure out what test to use. When we compare a mean to a known value, we perform a one-sample t-test. Specifcally, we are doing an unpaired one sample t-test, because our patient data is not linked to a previous or connected datapoint. We also call this a independent t-test.

First, let’s calculate it by hand. It is simply the (mean of your observations) minus the expected value, divided by the standard error of the mean.

hematocrit <- sickle$Hematocrit # i named my hematocrit variable for easier use
# do the t-test by hand
calc_ttest <- (mean(hematocrit)-40)/(sd(hematocrit)/sqrt(6))
calc_ttest
## [1] -8.740651

As you might expect, there is also a t-test function in R. The t.test() function is simple. See below.

t.test(hematocrit, mu=40)
## 
##  One Sample t-test
## 
## data:  hematocrit
## t = -8.7407, df = 5, p-value = 0.0003247
## alternative hypothesis: true mean is not equal to 40
## 95 percent confidence interval:
##  27.49041 33.17625
## sample estimates:
## mean of x 
##  30.33333

Part 1b.

As mentioned above, some t-tests are unpaired, or independent. However, what if we were measuring before an after treatment values, where the a pair of data come from the same person. Or, another example might be examining cells at different timepoints after the start of treatment, to see if longer duration of drug effects some measurable value. These are called paired tests, or dependent t- tests.

To run a paired t-test in R, it is very similar. We give the two values to compare, but then we tell R that this data is paired. It is important that the numbers line up, or to use a dataset file that has subjects or samples by rows.

hematocrit <- sickle$Hematocrit
hematocrit_baseline <- c(25, 27, 26, 30, 24, 24) # this is fake data
t.test(hematocrit, hematocrit_baseline, paired=TRUE)
## 
##  Paired t-test
## 
## data:  hematocrit and hematocrit_baseline
## t = 3.8762, df = 5, p-value = 0.01169
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  1.459591 7.207076
## sample estimates:
## mean of the differences 
##                4.333333

Part 2

Scientists were excited by being able to manipulate genes, and possibly treat a disease. With the discovery of CRISPR technology in the early 2000s, we might have found a way to permanantly affect genes. Scientists targeted the promoter region of gamma globin, where BCL11A binds. If they can change the sequence of the promoter, maybe we can permanantly elevate the production of gamma globin and fetal hemoglobin (HbF).

First, scientists tried this in mice. They had wildtype (WT) and mutant mice (targeting the bcl11a binding site). They collected total hemoglobin levels (HbTotal) as well as the percent of all hemoglobins that were of the gamma type (gammaglobin), expressed in percent. Here is some of the data.

# read the file; this is preloaded into your working directory
crispr <- read.csv("SickleCell/BIO205sickle2.csv", header = T)
crispr
##    Animal   HbTotal gammaglobin
## 1      WT 12.685470        0.02
## 2      WT 12.261350        0.02
## 3      WT 12.224960        0.01
## 4      WT 13.137680        0.03
## 5      WT 11.517520        0.02
## 6  bcl11a 10.204593       10.00
## 7  bcl11a 11.375821        9.50
## 8  bcl11a 12.588973        9.80
## 9  bcl11a  7.311909        7.50
## 10 bcl11a 12.168192        5.00

Now, let’s do some statistical testing. We want to

t.test(crispr$HbTotal~crispr$Animal)
## 
##  Welch Two Sample t-test
## 
## data:  crispr$HbTotal by crispr$Animal
## t = -1.6628, df = 4.6431, p-value = 0.1617
## alternative hypothesis: true difference in means between group bcl11a and group WT is not equal to 0
## 95 percent confidence interval:
##  -4.2234934  0.9524966
## sample estimates:
## mean in group bcl11a     mean in group WT 
##              10.7299              12.3654
bartlett.test(crispr$HbTotal ~ crispr$Animal)
## 
##  Bartlett test of homogeneity of variances
## 
## data:  crispr$HbTotal by crispr$Animal
## Bartlett's K-squared = 4.5644, df = 1, p-value = 0.03264
t.test(crispr$HbTotal~crispr$Animal, var.equal = T)
## 
##  Two Sample t-test
## 
## data:  crispr$HbTotal by crispr$Animal
## t = -1.6628, df = 8, p-value = 0.1349
## alternative hypothesis: true difference in means between group bcl11a and group WT is not equal to 0
## 95 percent confidence interval:
##  -3.9036602  0.6326634
## sample estimates:
## mean in group bcl11a     mean in group WT 
##              10.7299              12.3654

Part 3. CRISPR in humans.

Scientists use the information, similar to what you have been presented, to alter the BCL11A gene. Again, the goal was to increase the gene expression of gamma hemoglobin. They took stem cells from the bone marrow of a patient, Victoria Gray, and CRISPR edited BCL11A to alter its function. They put these cells back into Victoria, and waited. They continued to collect data on her hemoglobin types, and calculate her F cells. Data are shown below.

Discuss the results of this paper with your group.

Minute paper reflection

Go to this website and read a comic. Then, go to my polleverywhere and…
• Read the comic. Reflect on CRISPR technology. Who are the affected parties? What are the considerations?
• What is one thing that made sense
• What is one thing that was confusing

This week’s meme, because Elliot loves Spiderman.