# 6.00 Problem Set 7
# Problem 1
# Name: Jane Lee
# Collaborators: John Doe
# Time:
#

import pylab

# The *_age lists hold the evolutionary distance in millions of years.
# The *_sub lists hold the number of substitution per 100 amino acids
#     in the protein sequences.
# Adopted from Dickerson, J Mol Evol 1: 26-45 (1971).

# cytochrome c 
cyt_age = [90.0,90.0,240.0,300.0,300.0,400.0,600.0,1200.0]
cyt_sub = [10.2,5.0,13.6,10.0,15.3,19.8,29.6,60.1]

# globins
glob_age = [90.0,400.0,500.0,780.0,890.0,900.0,940.0,970.0,1000.0]
glob_sub = [16.3,68.9,87.0,138.6,151.4,156.5,161.9,163.5,174.2]

# fibrinopeptides
fib_age = [33.0,45.0,50.0,60.0,90.0]
fib_sub = [37.1,40.5,47.1,58.0,80.3]

def linreg_and_plot(xdata,ydata,data_marker,line_marker):
    """
    Fit y data against the x data using first order linear regression.
    Plot y data against x data with the specified data marker. Plot the fitted
    line with the specified line marker for x ranging from 0 to the maximum
    value of the x data. Return the coefficients of the fitted line.

    xdata,ydata: lists of numerical values, with the same number of elements
    data_marker: marker for the data points, as accepted by pylab.plot
    line_marker: marker for the fitted line, as accepted by pylab.plot
    returns: the coefficients of the fitted line
    """
    print 'linreg_and_plot not yet implemented.' # delete this line when you're done

def test_linreg_and_plot():
    # data for testing the regression and plotting helper funcion
    x_test = [1,2,3,4,5]
    y_test = [2,4,6,8,10]
    data_marker_test = 'xr'
    line_marker_test = '-r'
    pylab.figure()
    linreg_and_plot(x_test,y_test,data_marker_test,line_marker_test)
    pylab.title('test_linreg_and_plot')

def mol_clock():
    pylab.figure()
    print 'mol_clock not yet implemented.' # delete this line when you're done

print 'PROBLEM 1 data loaded'
print 'Cytochrome c:',len(cyt_age),'time points',len(cyt_sub),'substitution values.'
print 'Globins:',len(glob_age),'time points,',len(glob_sub),'substitution values.'
print 'Fibrinopeptides:',len(fib_age),'time points,',len(fib_sub),'substitution values.'

test_linreg_and_plot()
mol_clock()

pylab.show()