Approaching Rosalind Problems

I’ve been working through the Rosalind problems over the last week. They are an interesting set of problems exploring the space between Biology and Computation. I decided to take notes while doing one problem. This blog post is a basic outline of my process for problem solving.

Step 1 - Read The Problem And Background

Always read the problem. I’ve gotten a number of problems wrong in programming challenges because I didn’t read the instructions thoroughly. Rosalind provides some background info which I’ve found helpful as well. It is especially good for learning the relevant biology vocabulary. For this blog post I did the problem Mendel’s First Law.

Step 2 - Verify That You Can Read The Data

All of the Rosalind problems supply a data file and I’ve fallen into a pattern of starting out by reading in the file to ensure I know what’s going on. For this problem the provided file is 2 2 2. Since all the entries are the same it would be hard to verify I was reading and interpreting the file correctly. Because of this I used 1 2 3 as my example file. Then I wrote the code to put each of these values in a well named variable.

dominant, hetero, recessive = File.read("iprb.txt").chomp.split(' ').map(&:to_i)

puts "Dominant #{dominant}"
puts "Heterozygous #{hetero}"
puts "Recessive #{recessive}

Step 3 - Ponder And Devise A Strategy

My first strategy was to calculate how many ways there were to have the at least one dominant allele. Here’s some basic thoughts on that: From the background given I know that if one parent has the dominant allele the the resulting organism must have the dominant allele. Probability of one parent have the dominant allele is

dominant     not-dominant    dominant
--------  + ------------- * ----------
 total         total         total - 1

Then I remembered that since there are fewer ways get two recessive alleles I should calculate that instead and then subtract the result from one. (Basic law of probability).

Math:

Probability of two recessive parents mating:

recessive    recessive - 1
---------  * -------------
 total        total - 1

Probability of two heterozygous parents mating:

heterozygous     heterozygous - 1
------------  * ------------------
  total           total - 1

Punnet square for heterozygous mating

|   | Y  | y  |
|---+----+----|
| Y | YY | Yy |
| y | Yy | yy |

Only 1/4 of those will be recessive so we take the probability of two heterozygous mating and multiply it by 1/4.

heterozygous     heterozygous - 1      1
------------  * ------------------ * -----
  total           total - 1            4

Finally the probability of a heterozygous and a recessive organism mating:

heterozygous     recessive        recessive     heterozygous
------------  * ------------  + ------------ * --------------
  total           total - 1       total          total - 1

Punnet square for heterozygous & recessive mating

|   | Y  | y  |
|---+----+----|
| y | Yy | yy |
| y | Yy | yy |

In this case half the offspring have two recessive alleles so multiply the probability by 1/2.

Step 4 - Write The Code

First thing I realized was that we need floats so I had to change the import code slightly. I changed the to_i to a to_f.

dominant, hetero, recessive = File.read("iprb.txt").chomp.split(' ').map(&:to_f)

All my calculations require the total so I do that once:

total = dominant + hetero + recessive

Now calculate the probability of two recessive organisms mating:

r_r = (recessive / total) * ((recessive - 1) / (total - 1))

Now heterozygous organisms mating:

h_h = (hetero / total) * ((hetero - 1) / (total - 1))

Now the hetero + recessive matings:

h_r = (hetero / total) * (recessive / (total - 1)) + (recessive / total) * (hetero / (total - 1))

Now I incorporate the fractions from the punnet squares:

recessive_total = r_r + h_h * 1/4 + h_r * 1/2

This is the probability of a recessive organism. The problem asked for the probability of a dominant organism so I take 1 - recessive_total

puts 1 - recessive_total

All together:

dominant, hetero, recessive = File.read("iprb.txt").chomp.split(' ').map(&:to_f)

total = dominant + hetero + recessive

r_r = (recessive / total) * ((recessive - 1) / (total - 1))
h_h = (hetero / total) * ((hetero - 1) / (total - 1))
h_r = (hetero / total) * (recessive / (total - 1)) +
  (recessive / total) * (hetero / (total - 1))

recessive_total = r_r + h_h * 1/4 + h_r * 1/2
puts 1 - recessive_total
1 - recessive_total

When all this code runs it gives 0.78333 which is the expected result.

Step 5 - Download The Real Data Set

Now I downloaded the real dataset, ran it through my code, and pasted the result into the text box.

Step 6 - Celebrate

I recommend celebrating with cookies.

Post-script

This problem can also be solved by simulating all possible matings and then calculating the percentage that have the dominant allele.

dominant, hetero, recessive = File.read("iprb.txt").chomp.split(' ').map(&:to_f)

def is_dominant? organism
  organism.include? "Y"
end

d_ary = Array.new(dominant) { ["Y", "Y"] }
h_ary = Array.new(hetero) { ["Y", "y"] }
r_ary = Array.new(recessive) { ["y", "y"] }

ary = d_ary + h_ary + r_ary

children = []

(0...ary.length).each do |i|
  (0...ary.length).each do |j|
    next if i == j
    p1 = ary[i]
    p2 = ary[j]
    children << [p1[0], p2[0]]
    children << [p1[0], p2[1]]
    children << [p1[1], p2[0]]
    children << [p1[1], p2[1]]
  end
end

d_children, r_children = children.partition { |o| is_dominant?(o) }

puts d_children.count.to_f / children.count

Aja Hammerly