生物信息学中离散数学方法.pptVIP

Visualization of Real DNA DataAndA Few Nice Math Therefrom Bailin Hao (郝柏林) T-Life Research Center, Fudan University Institute of Theoretical Physics, Academia Sinica Santa Fe Institute /~hao/ / Not much biology A little statistics: Poisson distribution, Markovian prediction A little discrete mathematics: combinatorics, graph theory, formal language theory Everything from real biological data Biology Produces Huge Amount of Data Genomic sequences and complete genomes Protein sequences Gene expression data, microarray, gene regulation, protein interaction and network, metabolic network, immune network We will only look at DNA and protein sequences Biological Symbolic Sequences DNA (RNA): 1D, directed, unbranching heteropolymers made of 4 kinds of bases (nucleotides). Length: 104 to 108 nt Proteins: 1D, directed, unbranching heteropolymers made of 20 kinds of amino acids. Length: 102 to 104 AA Genome Projects World Wide(29 June 2008) Published：827（700 prokaryotes） On-going prokaryotes：90A + 1842B On-going eukaryotes: 936 On-going metagenomes：130 Total：3825 GenBank Rel. 166(15 June 2008) Sequences: 88 554 578 Nucleotides: 92 008 611 867 Average length: 1039 (almost the same in the last 8 years) Necessity to Look at Real Data Theorem 3 in Shannon’s seminal 1948 paper that laid the foundation of modern information theory: an intuitive explanation. Among 4N possible sequences of length N made of 4 kinds of letters: a huge subset of typical sequences and a tiny subset of atypical sequences. Biological sequences resulted from billions years of evolution and natural selection. They must belong to the atypical subset. They should be studied almost individually. A Simple-Minded Approach:Counting the Number of K-strings Different names: K-words, K-grams, K-strings, K-tuples The E. coli (strain K12) genome: a loop made of 4639475 letters Take K=8, there are 48=65536 string types. Do they all appear? If random, each string typ