BIOL 114forInweek

Discussion class worksheet

of Feb 11-15

GENETICS I: MENDEL’S LAWS AND PROBABILITY

Objectives:
 to work with and understand the basic mathematics of probability
 to work with and understand Mendel’s Laws

Introduction
This week, we begin our work in genetics by working with the basic mathematics of
probability, and by consulting Mendel’s Laws. From our work in class, you know that Mendel’s
fundamental insight was that inheritance is particulate. That is, genetically determined traits are
inherited as indivisible units that are not diluted from one generation to the next. Prior to this,
many biologists believed that traits “blended” from one generation to the next, and this
terminology still persists in modern speech such as “blending bloodlines” in human families or in
championship horses or dogs. Instead, Mendel’s Laws clearly spelled out the fact that inheritance
of many traits involves all-or-none transmission governed by the laws of probability.

Mendel’s First Law - The Law of Segregation - states that in diploid organisms two
alternate alleles segregate from each other in the germplasm and are passed separately without
dilution in gametes to the next generation.

Mendel’s Second Law - The Law of Independent Assortment - states that alleles of one
gene are passed to the next generation independently of alleles of another gene, and that the
process obeys the rules of random probability. A mathematical graph called a Punnet Square
allows one to determine the probabilities for each randomly generated genotype in the next
generation.

Although there are situations where one or both of Mendel’s Laws do not apply, a surprising
number of traits display Mendelian inheritance. You need to become thoroughly familiar with
the basics of Mendelian inheritance, and to be able two work simple problems derived from the
study of real cases.

We will first look at the basics of mathematical probability. Through thought problems and
simple experiments flipping coins or rolling dice, it’s easy to see how calculations work. It is
important that you become proficient in calculating probabilities, and in determining situations
where probabilities can be added or multiplied for more complex results.

In Class
Complete the following worksheets on probabilities and Punnett squares. If you are
unable to finish the data analyses during class time, finish them on your own time and bring them
to class next week.
 Genetics I In-Class Worksheet
To be completed by each student – due next week in discussion:

The Basics of Probability

Mathematical probability is a measure of the frequency of occurrence, or likeliness, of a given, or
desired, event. When the measure is used to describe events that have already occurred or pertain
to a population, the term “frequency” is commonly used. Most typically, the measure is used to
predict future events such as determining the “odds” in dice, cards, and other games of chance as
well as many other everyday situations, like the weather forecast and the likelihood of injury in a
car accident if you don’t wear a seatbelt. In these cases the word “probability” is most commonly
used. The basic formula for probability (p) is:
p= # of defined (or desired) events
# total events

As seen in the equation above, probability can be described as a number between 0 (if the # of
desired events = 0) and 1 (if the # of desired events = the # of total events). In these cases, p
equaling 0.0 indicates impossibility and p equaling 1.0 represents certainty.

In the case of predicting the future of events, the design of the system will often dictate the
probability of particular events occurring. For example, the probability of flipping a coin and
getting one particular side to be facing up is p = 0.5. This probability is based on the fact that the
coin has only two sides and should be fairly balanced through its center, allowing for equal
probabilities of one side vs. the other ending up showing. Based on the design of the coin, there
is no reason to suspect that one side of the coin is more likely to show one side over the other.
For systems that are more complex (eg. populations) probabilities are often determined by
previous calculations (frequencies). For example, our ability to predict that nearly half of the
nestlings of song sparrows (Melospiza melodia), a common small songbird, will not survive to
leave the nest is based on previous records of the species in the area where nestlings were taken
by predators at a frequency of 0.5.

Probabilities can be represented in numerous forms including decimals, fractions, percentages,

and ratios. We will use “p” to describe “probability of.”

Probability can be expressed as a:

- fraction - p = number of favorable events/total number of possible events. For example 1/3 is
read as, “one desired event out of a total of 3 events”,

- decimal – Completing the math of the fraction described above (ie dividing), results in a number
represented as a decimal, with values less than one (why is this so?). From the above fraction 1/3
= 1 divided by 3 = 0.333.
- percent - In order to find the percentage chance of an event occurring, one can convert its
decimal value to a percent by multiplying the decimal by 100. The 0.333 above multiplied by
100 yields a percentage of 33.3%.

- ratio - Number of favorable outcomes compared to the number of unfavorable outcomes. This
description is different than those described above, in that the total number of events is reflected
in the two numbers added together (in the above 3 examples it was read as the denominator of
the fraction, or as 1.0 for the decimal, or 100% for the percentage formats). The fraction 1/3
described above is described as a ratio by 1:2 and read as, “one desired event to two other
events.”

Complete the following thought experiments:

1. What is the chance of getting a head when tossing a coin? (give answer in all the above forms).

Fraction Decimal 0.5 Percent 501 Ratio 1 1

2. If you throw a 6-sided die (singular for dice), what is the chance of getting a four? (again give
answer in all forms).

Fraction Decimal 0.1667 Percent 16.671 Ratio 1 6

3. If you draw a card from a deck of 52 cards, what is the chance that you will get an ace?

Fraction Percent 7.691

Independence vs. Dependence
Considering multiple events is a bit trickier but can be easily mastered if we get a couple more
concepts out of the way.

Mutually exclusive events (those that are dependent) are events that cannot happen at the same
time. For example, an individual cannot get a head and a tail during a single coin flip. This can
be described in more general terms:
p(tail and head) = 0
The joint probability of mutually exclusive events occurring is their individual probabilities
added together. Therefore, the probably of getting heads or tails, is equal to 1. This can also be
described in more general terms:
p(head or tail) = p(head) + p(tail)
p(head or tail) = 0.5 + 0.5 = 1.0
In other words, if you flip a coin one time, you would expect a head or a tail to show up 100% of
the time.

Independent events are those that are not affected by previous events, nor do the events affect
each other. For example, if you were to flip two successive coins, the results from one toss
should not affect the results of the other toss. The probability of two independent events
occurring at the same time is their two independent probabilities multiplied together. In the case
of two coins being flipped, the probability of getting a head on one coin is p = 0.5, which is the
same as the probability of getting a head on the second coin as well. The probability of getting a
head on both coins, p(head, head), is 0.5 x 0.5 = 0.25. Similarly, the p(tail, tail) = 0.25. But
what about the remaining 0.5?

The other possibility of outcomes for our coin flipping example is to get one head and one tail in
the successive flips. However, there are two possibilities here: getting a head on the first coin
and tail on the second, AND getting a tail on the first coin and head on the second. Since the
resulting distribution is the same for both conditions (one head and one tail), we would like to
combine them as a single, mixed result. However, the two results cannot happen at the same
time; getting a head on the first coin automatically precludes the opportunity to get a tail on that
coin. Therefore, the two events are mutually exclusive, and their combined probabilities are
added together. Therefore, we describe p(head, tail or tail, head) = 0.5 (that is, 0.25 + 0.25 =
0.5). Notice that the three probabilities (head, head; head, tail; and tail, tail) added together
equals 1.0

4. If you throw a six-sided die, what is the chance you will get a 3 or a 5?

fraction Percent 33.33

5. What is the probability of getting a 1, a 2, a 3, a 4, a 5, or a 6? Show your calculations.

Eachnumber has a probability of 1 6
probability 100
f It to to to to 1
6. If you roll two dice, what is the probability that the first will be a six and the second will be a
three? Probability
offirstsix
6 2.781
Punnett Squares probability of second 3
I
Sometimes figuring out problems of joint probabilities is most easily done using Punnett squares.
A Punnett Square is a chart that predicts all possible numeric (genetic) combinations in a cross of
events (coin tosses, parents creating offspring, etc.) whose possible outcomes (coin side, allelic
combinations, etc.) are known. They are often used by genetic counselors, in order to find the
likelihood that a particular trait will be passed to an offspring and in the construction of Pedigree
Trees. Punnett squares are performed by utilizing Mendel’s Laws.

As a reminder:

Mendel’s Law of Segregation: In diploid organisms alleles from the same gene separate from
each other during meiosis and are passed into the gametes such that each gamete only gets one
allele from each gene.
Mendel’s Law of Independent Assortment: Alleles of one gene are passed to the gametes
independently of the alleles of other genes, and that the process obeys the rules of random
probability.

The mathematical table called a Punnett Square allows one to determine the probabilities for
each randomly generated genotype in the next generation by using the laws of probably we
discussed above.

For example, two parents have the genotype Dd at a particular locus. What are the different
allelic combinations for each gamete produced (that is, how many genetically distinct gametes
can be produced for each parent, and what are their genotypes), and what is the genotypic
distribution of the offspring for these parents?

First note, each genotype is described using two alleles (D or d). In diploid organisms (the most
common case for sexually reproducing organisms) there are two alleles for every gene. This is
because each chromosome includes a single allele for each gene, and there are two homologous
chromosomes (one from the female parent and one from the male parent) containing each gene
in every cell of the organism. Keep in mind we are ignoring sex chromosomes, conditions of
aneuploidy, allele deletions, etc., for the moment. Also note, in this example, there are only two
alleles (D and d), and they may occur in any combination of two (DD, Dd, or dd). There may be
several other allele types at this locus for this population, but we are only working with two here.

So, parent #1 has genotype Dd and #2 has Dd. Given Mendel’s Law of Segregation, alleles
separate and go independently into separate gametes. Therefore, we would say that each gamete
(column and row titles in the Punnett Square below) has only one of each possible allele. For
example, the first gamete receives D and the second receives d. Parent #2 does the same thing,
since they have the same genotype as parent #1.

Parent 1
D d
D
Parent 2 DD Dd
d
Dd dd

We can now fill in the squares by combining gametes from parents 1 and 2 in all combinations.

Parent 1
D d
D DD Dd
Parent 2
d
Dd dd

The four resulting genotypes represent the range of possibilities the offspring may have for these
two parents for this particular gene. Note, although the parents had identical genotypes for this
trait, only half of their potential offspring will be like them genetically (Dd and dD). One quarter
will be homozygous dominant (DD) and the last quarter will be homozygous recessive (dd).

Now let’s expand on this to include actual probabilities. The example of the two coin flip is
described in the following Punnett Square:

Coin 1
½H ½T
½ H ¼HH ¼HT
Coin 2
½T
¼TH ¼TT

There are a few things to note about how this Punnett Square is set up. With Coin 1, the two
possible outcomes are heads (H) and tails (T), and each has a probability of ½ of occurring. In
terms of Mendel’s laws, the coin represents the gene and the gene has two possible alleles (H and
T). Also in this case, the gametes formed during meiosis only have a single gene (coin side).
Coin 2 has the same possibilities as Coin 1. If each column and row heading represents a
gamete, the information contained inside the cell of intersection between each row and column
represents the possible outcome (offspring) of gametes for that row and column. For example, if
Coin 1 provides gamete 1/2H, and Coin 2 provides 1/2H, the resulting probability of getting HH
across the two coins is 1/4HH. In other words, of all the possible outcomes of this two-coin flip,
25% of the time we would expect the result to be HH. The values of the gametic probabilities
are multiplied together (their individual probabilities are independent of each other). Also notice
that the probability of getting offspring that are a mix of the two types of gametes is 1/2 (or 1/4 +
1/4).

7. Imagine standing several feet away from 3 equally-sized garbage cans (A, B, and C) that are
all pushed together such that they are all touching each other. Imagine also, that you want to
throw 2 wadded up pieces of paper into the garbage cans, but the aerodynamics of the paper balls
are such that the paper curves unpredictably in the air as it travels to the cans. Repeated testing
concludes that the paper lands in any one of the cans completely randomly. What is the
probability that both paper balls will land in can B? How about cans A and C? How about B
and C? Use a Punnett Square with associated probabilities to form your answer.

A B C
probability bothlanding in B

A AA AB AC
Probability landing A QC

B BA BB BC

c CA CB C Probability
landingBQC 5 3
8. What would happen if can C has an opening 2x the area of either of the other cans? What is
the probability of the two pieces of paper landing both in B, A and C, B and C? Again, use a
Punnett Square to determine your answers.

A B Probability of B QA
C landing 2x
4
A AA AB AC Probability of landing CQA 2x

B BA BB BC
Probability of landing in both
C CA CB CC

9. Up until now, both events (or parental gametic possibilities) have been equal. However, they
do not have to be. How would the scenario in #8 above have changed if during the tossing of the
second paper ball, can C was removed. Assuming you would still get the paper into one of the
cans, what is the probability of the paper balls landing in cans A and A, A and B, A and C? Use
a Punnett Square to determine the probabilities. Keep in mind that each of the column and row
titles (gametes) must contain one potential outcome (allele) from each event.

A B C probability of landing in both AQA

AA AB AC
A
4 1 0 f
BA BB BC
B Probability of landing in both A B
4 12 0
2x 4 t
c

Probability of landing in both AQC

E
 10. What is the range of potential genotypes of a parental cross of Dd and dd?

Dd and dd

11. Obtain a coin. Toss it in the air, catch it, then place it on your wrist with your hand over it.
Without looking at the coin, calculate the probability that it will be heads up this time. Toss the
coin again. On the basis of the previous toss, what is the probability that the coin will be heads this
second time?

1st toss 501 heads 2nd toss 501

If you tossed the coin 5 times and got all heads, would you really be surprised by the results?
What’s the probability? Would you suspect your results to be invalid? For instance, would you
suspect that the coin is weighted?
The probability would be 3.125 of geting all heads Yes I would be really shocked
invalid but
because it is rare to
get all headsand a low percentage The probability is not
very rare andlucky Thecoinisprobably weightedtowards heads because our outcome isextremely
Now toss your coin 20 times and record the results in the chart below. Compare your results with rare
tosses made by others in your group, then fill out the chart below.

Expected Your 20 tosses Your group’s total

heads 10 12 44
tails 10 8 56
Total 20 20 100
12. For each of the above, how closely do the results agree with your expectations?

Myresults are different by a little because I expected it to be 10 10

The probability of it turningheads was a littlemore than expected
13. In which of the above would you expect the results to most nearly approach your
expectation? Why?
I expect the grouptotal tomostnearly approachmy expectations because
the biggerthe samplesize the higherpossibility of it reaching my expectations of halt half
14. What can you conclude about the relationship between sample size (the number of tosses in
this case) and closeness of the result to the expected ratio?

The higher thesamplesize the higher chanceof probabilityof it reaching

expected 501 heads and 50t tails more closely
Statistical Analysis of Genetic Data

The ratio one actually observes when flipping a coin or counting corn kernels is typically
not exactly the probability one might predict. For instance, flipping a coin a number of times
often yields not quite equal numbers of heads and tails even though we predict a ratio of 1:1.
Why do you think this is the case?

In judging the results of real data compared with hypothetical predictions, if the deviation
- the difference between your actual and expected results- is small, one is rarely surprised. By
experience we are intuitively prepared for some deviation in any random process. Suppose,
however, that you end up with a big difference. In that event, it’s time to look for some cause for
why actual results fail to meet expectations. Perhaps the experiment was run poorly (large error
or data collection bias), or maybe something’s wrong with what you expected. For instance, you
may have an unbalanced coin If so, then the expectations are different.

In many situations, it’s unclear whether the deviation is sufficiently large to be of concern
or not. Just how large must a deviation become before one decides that the data don’t fit
expectations? In many scientific studies, such as ascertaining risk of some drug in cancer
patients, or in a field study involving relationships between pollinator and pollinated plants, one
simply cannot deal with this problem by guessing whether the results are close enough. For these
(and many other) cases, the concept of statistical significance is employed.

The first step of any analysis is to specify the hypothesis to be tested. In this, as in most
statistical tests, we are testing a Null Hypothesis (H0). Basically, the null hypothesis is a
statement of "no difference" or "randomness". If things are happening by chance or randomness,
you would expect to see no differences, on average, amongst your data. Keep in mind that the
null hypothesis may not directly reflect what you are expecting as a conclusion to your research.
You may very well expect to find differences in your data, but the idea of "no difference" is
tested so that we may eliminate it as a possibility. REMEMBER THAT SCIENCE IS A
PROCESS OF ELIMINATION. SCIENTISTS DO NOT PROVE THINGS; THEY REJECT
IDEAS AND SUPPORT OTHERS -- NEVER PROVE.

Once a null hypothesis is established, an Alternative Hypothesis (H1) is implied. The

alternative hypothesis is generally the opposite of the null (i.e., there is a difference). Sometimes
the alternative hypothesis can be more specific (depending on the test you are performing), but it
does not need to be.

For these analyses we will be using the chi-square goodness of fit test, which is meant to
examine the fit of experimental data to a hypothetical ratio. The underlying idea is that
experimentally derived ratios, even if they correspond to an expected ratio are likely to deviate a
little from that ratio just due to chance. But if they deviate to the extent that there is a less than 5
% chance (the usual figure, but some prefer 1%) of getting that data set by chance, then the
experimental data is usually considered unlikely to represent the expected ratio, i.e., the "real"
expected ratio is something different. To illustrate, consider the following problem.
 Suppose you suspect that a "friend" has given you a weighted die to toss. You must first
test the null hypothesis that the die is fair. So, you toss the die 90 times. If the die isn't weighted,
you would expect to get a "1" 15 times, a "2" 15 times, etc. to "6". In fact, however, you get "6"
7 times, "5" 15 times, "4" 12 times, "3" 18 times, "2" 13 times and "1" 25 times. The results look
suspiciously weighted toward "1" and away from "6". You can use the chi-square test to
determine whether the observed deviation from the expectation of a fair die is too large.

Here’s how:

"1" "2" "3" "4" "5" "6" SUM

1) Write down your data &
25 13 18 12 15 7 90
sum
2) Write expected numbers
15 15 15 15 15 15 90
for same sum
3) Note the deviations +10 -2 +3 -3 0 -8
4) Square the deviations 100 4 9 9 0 64
5) Divide squared deviations
100/15 4/15 9/15 9/15 0/15 64/15
by the expected ratio:

6) Sum the last row = 186/15 = 12.4 = Chi-square Value (Note: the steps 1-6 are really a
representation of the chi-square formula X2= ∑ (obs-exp)2
exp
7) Calculate degrees of freedom = number of categories (6 possible answers) minus one = 5.

8) Construct a decision rule. In theory, at least, any observed data might agree with the null
hypothesis, but those showing greater deviation from expected ratios are less likely. Many
scientific studies set a boundary probability at 1/20 or 5% (0.05) and therefore construct the
following decision rule:

Reject the null hypothesis if the probability that the observed data fits the null hypothesis is
less than 5%.

9) Use the chi-square table below (derived from extensive statistical calculations) to calculate the
probability for your data. From the chart, we see that for 5 degrees of freedom, a chi-square
value of 12.4 yields a probability of between 0.05 and 0.01.

Because the chi-square value lies below a probability of 0.05, you may conclude that the
die is, in fact weighted. However, suppose you require a higher standard of “proof” that your
"friend" is a cheat - i.e., "innocent until proven guilty"! You might choose to give your “friend”
the benefit of the doubt and insist instead on a decision rule with boundary set at 1%. In this
case, the probability estimated from the chi-square chart is not sufficiently low to reject the null
hypothesis of a fair die. Also, if your friendship depends on the result, you might want to collect
more data since only 90 trials seems a small sample on which to base such a momentous
decision. Notice here that decisions made with statistics should never be viewed as providing
absolute proof. At this point, other factors enter in. Perhaps the die is fair, but the way you throw
it, perhaps unconsciously, introduces a systematic bias in the result. Another thrower might
obtain a less biased result.

Values of Chi-square
Probabilities:

P 0.99 0.9 0.8 0.5 0.2 0.1 0.05 0.01

df
1 0.00 0.02 0.06 0.46 1.64 2.71 3.84 6.64
2 0.02 0.21 0.45 1.39 3.22 4.61 5.99 9.21
3 0.12 0.58 1.00 2.37 4.64 6.25 7.82 11.35
4 0.30 1.06 1.65 3.36 5.99 7.78 9.49 13.28
5 0.55 1.61 2.34 4.35 7.29 9.24 11.07 15.09
6 0.87 2.20 3.07 5.35 8.56 10.65 12.59 16.81
7 1.24 2.83 3.82 6.35 9.80 12.02 14.07 18.48
8 1.65 3.49 4.59 7.34 11.03 13.36 15.51 20.09
9 2.09 4.17 5.38 8.34 12.24 14.68 16.92 21.67
10 2.56 4.87 6.18 9.34 13.44 15.99 18.31 23.21
15 5.23 8.55 10.31 14.34 19.31 22.31 25.00 30.58
20 8.26 12.44 14.58 19.34 25.04 28.41 31.41 37.57
25 11.52 16.47 18.94 23.34 30.68 34.38 37.65 44.31
30 14.95 20.60 23.36 29.34 36.25 40.26 43.77 50.89
Deviations significant.
Deviations from expected are insignificant
Good reason to doubt
Therefore, no reason to doubt null hypothesis.
the null hypothesis.
 Worked with Kelly Hom

14. Complete a Χ2 Goodness of Fit Test for your coin data from the table above. Be sure to
include a null hypothesis, alternative hypothesis, table with observed and expected values, total
Χ2, degrees of freedom, statistical conclusion and overall conclusion.

44 heads
56 tails

Null Hypothesis The probability of getting heads is equal to the probability of gettingtails

Alternative hypothesis The probability of getting heads is notequal to the probabilityof gettingtails

heads tails sum

observed 44 56 100

expected 50 50 100

deviation 6 6
squared
deviation 36 36
2
on 1.44 total
dividedby 5
expected

Degree of freedom 2 1 1

statistical conclusion

There is a 5 probability at 1 degree of

2
freedom which is 3.84 of makingthe total 1.44
less than 3.84

overall conclusion

Based on the chi square of fit test there

goodness
the nullhypothesis
is not enoughevidence to prove to reject
The total x2 value is below the probability of 0.05
One important biological trait for expecting parents is the sex of their offspring. Answer the
following questions:

15. If a couple has one baby, what is the probability that the child will be boy?

501
16. If they have two children, what is the probability that both will be girls?

1 1 251
17. If they have four children, what is the probability that all will be boys?

4 6.25 t
18. If they have three girls, what is the probability that the fourth child will also be a girl?

501
19. If they have six girls, what is the probability that the seventh child will also be a girl?

501
The Gambler’s Fallacy, or the fallacy of the maturity of chances, is the mistaken belief that if
something happens more frequently than what is considered normal, then it will happen less
frequently than normal in the future, even if the observer knows that the independent probability of
that event occurring is already established. In other words, it is incorrect to believe that, due to
some universal equilibrium, if the same event occurs repeatedly, then you should expect that it is
less likely to occur in the near future and the opposite event will occur to balance out the series of
events.

The reverse of the fallacy can also occur, but may be more difficult to define. If an event occurs
repeatedly when the laws of probability suggest outcomes should be random, one must determine
whether those repeated occurrences are simply a string of similar outcomes occurring by chance, or
if the system is biased in some way to provide an increased likelihood of gaining that one result.
Most people are likely to assume the system is biased toward that one result, even when they know
the potential outcomes are equally probable before they consider the question.