1. Basic Statistics and Shannon Entropy Ka-Lok Ng Asia University

2. Mean and Standard Deviation (SD) Compare distributions having the same mean value –a small SD value  a narrow distribution –a large SD value  a wide distribution

3. Pearson correlation coefficient or Covariance Statistics – standard deviation and variance, var(X)=s 2, for 1-dimension data How about higher dimension data ? - It is useful to have a similar measure to find out how much the dimensions vary from the mean with respect to each other. - Covariance is measured between 2 dimensions, - suppose one have a 3-dimension data set (X,Y,Z), then one can calculate Cov(X,Y), Cov(X,Z) and Cov(Y,Z) - to compare heterogenous pairs of variables, define the correlation coefficient or Pearson correlation coefficient, -1 ≦  XY ≦ 1 -1  perfect anticorrelation 0  independent +1  perfect correlation

4. The squared Pearson correlation coefficient Pearson correlation coefficient is useful for examining correlations in the data One may imagine an instance, for example, in which an event can cause both enhancement and repression. A better alternative is the squared Pearson correlation coefficient (pcc), The square pcc takes the values in the range 0 ≦  sq ≦ 1. 0  uncorrelate vector 1  perfectly correlated or anti-correlated pcc are measures of similarity Similarity and distance have a reciprocal relationship similarity↑  distance↓  d = 1 –  is typically used as a measure of distance

5. Pearson correlation coefficient or Covariance - The resulting  XY value will be larger than 0 if a and b tend to increase together, below 0 if they tend to decrease together, and 0 if they are independent. Remark:  XY only test whether there is a linear dependence, Y=aX+b - if two variables independent  low  XY, - a low  XY may or may not  independent, it may be a non-linear relation, for example, y=sin x) - a high  XY is a sufficient but not necessary condition for variable dependence

6. Pearson correlation coefficient To test for a non-linear relation among the data, one could make a transformation by variables substitution Suppose one wants to test the relation u(v) = av n Take logarithm on both sides log u = log a + n log v Set Y = log u, b = log a, and X = log v  a linear relation, Y = b + nX  log u correlates (n>0) or anti-correlates (n<0) with log v

7. Pearson correlation coefficient or Covariance matrix A covariance matrix is merely collection of many covariances in the form of a d x d matrix:

8. Spearman’s rank correlation One of the problems with using the PCC is that it is susceptible to being skewed by outliers: a single data point can result in situation appearing to be correlated, even when all the other data points suggest that they are not. Spearman’s rank correlation (SRC) is a non-parametric measure of correlation that is robust to outliers. SRC is a measure that ignores the magnitude of the changes. The idea of the rank correlation is to transform the original values into ranks, and then to compute the correlation between the series of ranks. First we order the values of gene A and B in ascending order, and assign the lowest value with rank 1. The SRC between A and B is defined as the PCC between ranked A and B. In case of ties assign midranks  both are ranked 5, then assign a rank of 5.5

9. Spearman’s rank correlation The SRC can be calculated by the following formula, where x i and y i denote the rank of the x and y respectively. An approximate formula in case of ties is given by

10. Distances in discretized space Sometimes one has to due with discreteized values The similarity between two discretized vectors can be measured by the notion of Shannon entropy.

11. Entropy and the Second Law of Thermodynamics: Disorder and the Unavailability of energy Ice melts, it becomes more disordered and less structured. Entropy always increase

12. Statistical Interpretation of Entropy and the Second Law S = k ln  S = entropy, k = Boltzmann constant, ln  = natural logarithm of the number of microstates  corresponding to the given macrostate. L. Boltzmann (1844-1906) http://automatix.physik.uos.de/~jgemmer/hintergrund_en.html

13. Entropy and the Second Law of Thermodynamics: Disorder and the Unavailability of energy

14. Concept of entropy Toss 5 coins, outcome 5H0T1 4H1T5 3H2T10 2H3T10 1H4T5 0H5T1 A total of 32 microstates. Propose entropy, S ~ no. of microstates,  i.e. S ~  Generate coin toss with Excel The most probable microstates

15. Shannon entropy Shannon entropy is related to physical entropy Shannon ask the question “What is information ?” Energy is defined as the capacity to do work, not the work itself. Work is a form of energy. Define information as the capacity to store and transmit meaning or knowledge, not the meaning or knowledge itself. For example, a lot of information from WWW, but it does not mean knowledge Shannon suggest entropy is the measure of this capacity Summary Define information  capacity to store knowlege  entropy is the measure  Shannon entropy Entropy ~ randomness ~ measure of capacity to store and transmit knowledge Reference: Gatlin L.L, Information theory and the living system, Columbia University Press, New York, 1972.

16. Shannon entropy How to relate randomness and measure of this capacity ? Microstates 5H0T1 4H1T5 3H2T10 2H3T10 1H4T5 0H5T1 Physical entropy S = k ln  Shannon entropy Assuming equal probability of each individual microstate, p i p i = 1/  S = - k ln p i Information ~ 1/p i =  If p i = 1  there is no information, because it means certainty If p i << 1  there are more information, that is information is a decrease in uncertainty

17. Distances in discretized space Sometimes it is advantageous to use a discretized expression matrix as the starting point, e.g. to assign values 0 (expression unchanged, 1 (expression increased) and -1 (expression decreased). The similarity between two discretized vectors can be measured by the notion of Shannon entropy. Shannon entropy, H 1 -Probability of observing a particular symbol or event, p i, with in a given sequence Consider a binary system, an element X has two states, 0 or 1 base 2 References: 1. http://www.cs.unm.edu/~patrik/networks/PSB99/genetutorial.pdf http://www.cs.unm.edu/~patrik/networks/PSB99/genetutorial.pdf 2. http://www.smi.stanford.edu/projects/helix/psb98/liang.pdfhttp://www.smi.stanford.edu/projects/helix/psb98/liang.pdf 3. plus.maths.org/issue23/ features/data/plus.maths.org/issue23/ features/data/ Claude Shannon - father of information theory - H 1 measure the “ uncertainty ” of a probability distribution - Expectation (average) value of information and

18. Shannon Entropy pipi 1,0-1*1[log 2 (1)] = 0 ½, ½-2*1/2[log 2 (1/2)]=1 2 2 -states 1/4-4*1/4[log 2 (1/4)]=2 2 N -states 1/2 N -2 N *1/2 N [log 2 (2 N )]=N Uniform probability certain No information Maximal value DNA seq.  n = 4 states, maximum H 1 = - 4*(1/4)* log(1/4) = 2 bits Protein seq.  n = 20 states,  maximum = - 20*(1/20)*log(1/20) = 4.322 bits, which is between 4 and 5 bits.

19. The Divergence from equi-probability When all letter are equi-probable, p i = 1/n H 1 = log 2 (n)  the maximum value H 1 can take Define H max 1 = log 2 (n) Define the divergence from this equi-probable state, D 1 D 1 = H max 1 - H 1 D 1 tells us how much of the total divergence from the maximum entropy state is due to the divergence of the base composition from a uniform distribution For example, E. coli genome has no divergence from equi-probability because H 1 Ec = 2 bits, but, for M. lysodeikticus genome, H 1 Ml = 1.87, then D 1 = 2.00 – 1.87 = 0.13 bit Divergence from independence Single-letter events  which contains no information about how these letters are arranged in a linear sequence H max 1 H 1 = log 2 (n) D1D1

20. Divergence from independence – Conditional Entropy Question Does the occurrence of any one base along the DNA seq. alter the probability of occurrence of the base next to it ?  What are the numerical values of the conditional probabilities ?  p(X|Y) = prob. of event X condition on event Y  p(A|A), p(T|A), p(C|A), p(T|A) … etc.  If they were independent, p(A|A) = p(A), p(T|A) = p(T) ….  Extreme ordering case, equi-probable seq., AAAA…TTTT…CCCC…GGGG…  p(A|A) is very high, p(T|A) is very low, p(C|A) = 0, p(G|A) = 0  Extreme case, ATCGATCGATCG….  Here p(T|A) = p(C|T) = p(G|C) = p(A|G) = 1, and all others are 0  Equi-probable state ≠ independent events

21. Divergence from independence – Conditional Entropy Consider the space of DNA dimers (nearest neighbor) S 2 = {AA, AT, …. TT} Entropy of S 2, H 2 = -[p(AA)log p(AA) + p(AT) log p(AT) + …. + p(TT) log(TT)] If the single letter events are independent, p(X|Y) = p(X), If the dimer event is independent, p(A|A)=p(A)p(A), p(A|T)=p(A)p(T), …. If the dimer is not independent, p(XY) = p(X)p(Y|X), such as p(AA) = p(A)p(A|A), p(AT) = p(A) p(T|A) … etc. H Inp 2 = entropy of completely independent Divergence from independence, D 2 = H Inp 2 – H 2 D 1 + D 2 = the total divergence from the maximum entropy state

22. Divergence from independence – Conditional Entropy Calculate D 1 and D 2 for M. phlei DNA, where p(A)=0.164, p(T)=0.162, p(C)=0.337, p(G)=0.337. H 1 = -(0.164 log 0.164 + 0.162 log 0.162 +..) = 1.910 bits D 1 = 2.000 – 1.910 = 0.090 bit See the Excel file D 2 = H Inp 2 – H 2 = 3.8216 – 3.7943 = 0.0273 bit Total divergence, D 1 + D 2 = 0.090 + 0.0273 = 0.1173 bit

23. Divergence from independence – Conditional Entropy - Compare different sequences using H to establish relationships -Given the knowledge of one sequence, say X, can we estimate the uncertainty of Y relative to X ? -Relation between X, Y, and the conditional entropy, H(X|Y) and H(Y|X) -conditional entropy is the uncertainty relative to known information H(X,Y) = H(Y|X) + H(X) = uncertainty of Y given knowledge of X, H(Y|X) + uncertainty of X, sum to the entropy of X and Y = H(X|Y) + H(Y) Y=2 x log 10 Y=xlog 10 2 X=log 10 Y/log 10 2 H(Y|X) = H(X,Y) – H(X) = 1.85 – 0.97 = 0.88 bit where

24. Shannon Entropy – Mutual Information Joint entropy H(X,Y) where p ij is the joint probability of finding x i and y j probability of finding (X,Y) p 00 = 0.1, p 01 = 0.3, p 10 = 0.4, p 11 = 0.2 Mutual information entropy, M(X,Y) Information shared by X and Y, or it can be used as a similarity measure between X and Y H(X,Y)= H(X) + H(Y) – M(X,Y)  like in set theory, A ∪ B = A + B – (A∩B) M(X,Y)= H(X) + H(Y) - H(X,Y) = H(X) – H(X|Y) = H(Y) – H(Y|X) = 1.00 – 0.88 M(X,Y)= H(X) + H(Y) – H(X,Y) = 0.97 + 1.00 – 1.85 = 0.12 bit and

25. Shannon Entropy – Conditional Entropy H(Y|X) f(Y|X=x) 0 0 1/4 1 0 3/4 0 1 4/6 1 1 2/6 Conditional entropy a particular x All x’s