DemoScriptSequence.py

##### Demonstration Script #####
import Sequence
from Sequence import Sequence as sq
import matplotlib.pyplot as plt


num_seq = int(input("Enter the number of sequences: ")) #Take the number of sequences the user wants to input
seq_list = []

#Create Multiple Class Instances
for i in range(0,num_seq):
	dna = sq(str(input("Enter the DNA Sequence: ")))
	print("*****CREATING NEW SEQUENCE INSTANCE*****")
	seq_list.append(dna)
	print("***Object Created***\n")


#Testing Base Count
len_seq = sq.base_count(seq_list[0])
print("Length of the sequence is ", len_seq)

# Testing is_dna
valid_dna = sq.is_dna(seq_list[0])
assert valid_dna == True, "This is not a valid DNA sequence"


#Testing __eq__
dna_1 = sq("ATGCTGCA")
dna_2 = sq("atgctgca")
dna_3 = sq("ATGATGCA")

true_val = (dna_1 == dna_2)  #The DNA Sequences are same
true_val2 = (dna_1 == dna_3) #The DNA Sequences are different

assert true_val == True, "The DNA sequences are not equal and not of the same object type"
assert true_val2 == False, "The sequences are equal" 


#Testing comp_seq
dna_input = sq("ATGCTGCT")
dna_comp = sq("TACGACGA")
compliment_seq = sq.comp_seq(dna_input)

if dna_comp == compliment_seq:  #Compare the complimentary sequence generated by the method with the manually defined complimentary sequence
	print("The complimentary sequences are the same \n \n")
else:
	print("ERROR: The complimentary sequences are not same \n \n")


#Testing MisMatch
mismatch_index = sq.mis_match(dna_1,dna_3)
print("Mismatch found at position: ",mismatch_index)
print("\n")


#Reading genome_01.dat
print("***READING genome_01.dat***")
genome_01 = Sequence.read_genome("genome_01.dat")


#Testing gene_finder
genes_1 = sq.gene_finder(genome_01)
gene_len1 = sq.base_count(genes_1[0])
print("The length of first gene is: ", gene_len1)

print("\n")

#Testing Swap Mutation
print("***READING genome_02.dat***")
genome_02 = Sequence.read_genome("genome_02.dat")
genes_2 = sq.gene_finder(genome_02)

gene_mut = []
gene_len = []

#Count number of mutation for each gene comparison
for i in range(0,len(genes_1)):
	mut_count = sq.swap_mut(genes_1[i],genes_2[i])  #Counts number of mutation for each gene comparison
	gene_mut.append(mut_count)
	gene_count = sq.base_count(genes_1[i])
	gene_len.append(gene_count)




#GeneLength Histogram
plt.style.use(['ggplot','dark_background'])
plt.figure(0,figsize = (19,9))
plt.title("Gene Length Bar")
plt.xlabel("Gene", fontsize = "15")
plt.ylabel("Gene Length", fontsize = "15")
plt.bar(range(len(gene_len)), gene_len)
plt.savefig("GeneLength_NairKarthik.png")

#Gene Mutation vs Gene Length
plt.figure(1,figsize = (19,9))
plt.title("Gene Mutation vs Gene Length", fontsize = "20")
plt.xlabel("Gene Mutation", fontsize = "15")
plt.ylabel("Gene Length", fontsize = "15")
plt.scatter(gene_mut,gene_len,color = "red")
plt.savefig("GeneMutVsLen_NairKarthik.png")
plt.show()