#!/usr/bin/env python
#Author: Shriphani Palakodety (spalakod@cs.purdue.edu)




from math import sqrt, factorial

def combs(num1, num2):
	'''Return the value of num1 choose num2'''
	return float(factorial(num1)) / (factorial(num1-num2) * factorial(num2))


class Protein:
	'''Describes the protein'''
	def __init__(self, vector):
		self.vector = vector #stores the vector representation of the protein's functions
	
	def __str__(self):
		return str(self.vector)

	def similarityMeasure(self, protein2):
		'''Takes a protein for input and determines the similarity between this protein and that'''
		similarity = 0
		for i in self.vector:
			if i in protein2.vector:
				similarity += 1
		return float(similarity)/(sqrt(len(self.vector)) * sqrt(len(protein2.vector)))


class ProteinSet:
	'''Describes a set of proteins'''
	def __init__(self):
		self.proteins = []
		self.coherence_scr = 0
		#self.setCoherenceScore()
		self.core = []
		self.similarityCritereon1(self)

	def __str__(self):
		print "List of proteins in this set: "
		for prot in self.proteins:
			print "-> ", prot
		return ""

	def addProtein(self, protein):
		'''Add a protein to this set'''
		self.proteins.append(protein)
		self.setCoherenceScore()
		self.similarityCritereon1(self)
	
	def removeProtein(self, protein):
		'''Remove a protein from this set'''
		self.proteins.pop(self.proteins.index(protein))
		self.setCoherenceScore()
		self.similarityCritereon1(self)

	def setCoherenceScore(self):
		'''Set the coherence score of the set'''
		self.coherence_scr = 0
		if len(self.proteins) <= 1:
			self.coherence_scr = 1
			return
		for i in xrange(len(self.proteins)):
			for j in xrange(i+1, len(self.proteins)):
				self.coherence_scr += self.proteins[i].similarityMeasure(self.proteins[j])
		self.coherence_scr =  2 * float(self.coherence_scr)/(len(self.proteins) * (len(self.proteins) - 1))
				
	
	def similarityMeasureSet(self, protein):
		'''Given a protein and a set, this function returns the similarity between the protein and the set'''
		similarity = 0
		set_prot_list = self.proteins
		for prot in set_prot_list:
			similarity += protein.similarityMeasure(prot)
		return float(similarity)/len(set_prot_list)

	def similarityCritereon1(self, set):
		'''Given a set, use the first criterion to figure out the similarity'''
		prots = 0
		if self.proteins == set.proteins:
			self.core = []
			for prot in set.proteins:
				if self.similarityMeasureSet(prot) >= self.coherence_scr:
					prots += 1
					self.core.append(prot)
		else:
			for prot in set.proteins:
				if self.similarityMeasureSet(prot) >= self.coherence_scr:
					prots += 1
					
		return prots
	
	def similarityCritereon2(self, set):
		'''Given a set, use the second criterion to figure out the similarity'''
		prots = 0
		for prot in set.proteins:
			for set_prot in self.core:
				print str(prot) + "," + str(set_prot)
				if set_prot.similarityMeasure(prot) >= self.coherence_scr:
					prots += 1
		return prots
	
	def similarityCritereon3(self, set):
		'''Given a set, use the third critereon to figure out the similarity'''
		prots = 0
		for prot in set.proteins:
			for set_prot in self.proteins:
				if set_prot.similarityMeasure(prot) >= self.coherence_scr:
					prots += 1
		return prots



def getPValue(num_prots, set1_num, set2_num):
	p_value = 0
	for i in xrange(set1_num):
		p_value += combs(num_prots) * combs((set2_num - num_prots), (set1_num -i)) / float(combs(set2_num, set1_num))
	return p_value



#Creating a bunch of proteins and so on for "fun" and "entertainment"

prot1 = Protein(['f1', 'f2', 'f3'])
prot2 = Protein(['f1', 'f2', 'f3'])


#print prot1.similarityMeasure(prot2)

S = ProteinSet()
S.addProtein(prot1)
S.addProtein(prot2)
print S.similarityMeasureSet(prot2)

#print S

#print S.coherence_scr
prot3 = Protein(['f1', 'f2'])

#compare the similarities between prot3 and the set S

#print S.similarityMeasureSet(prot3)


#here onwards we begin implementing the core and so on.

print len(S.core)

for prot in S.core:
	print prot

R = ProteinSet()
R.addProtein(prot3)

prot4 = Protein(['f1', 'f2', 'f3'])

prot5 = Protein(['f1', 'f5', 'f9'])
prot6 = Protein(['f1', 'f3', 'f7'])

R.addProtein(prot4)
R.addProtein(prot5)
R.addProtein(prot6)

print R

print S.similarityCritereon1(R)
print S.similarityCritereon2(R)
print S.similarityCritereon3(R)