November 22, 2021 [reading notes] - bioinformatics and functional genomics (Chapter 5, section 4, hidden Markov model)

• 5.4 Spectral search ： hidden Markov model （HMMs）
  • The spectral hidden Markov model is used to generate a position specific scoring system for identifying the similarity of distant sequences , Than PSSMs A more general , Such as voice detection , A series of signal detection problems such as sonar ;
  • In the field of bioscience ,HMMs It has been used in a wide variety of applications ： Sequence alignment 、 Protein structure prediction 、 Protein transmembrane region prediction 、 Chromosome copy number change analysis and gene discovery algorithm ;
  • advantage ：
    • Spectrum HMMs It's a probability model , It evaluates the occurrence of a match at a given location in the comparison 、 Mismatch 、 Insert and confirm （ vacancy ） The possibility of . By developing a statistical model based on known sequences , We can use spectrum HMM To describe the possibility of a particular sequence matching the model .
    • Spectrum HMM A multiple sequence alignment can be transformed into a position specific scoring system . Spectrum HMM A common application of is in spectrum HMMs Query a single protein sequence of interest in a database , In another application, a spectrum is used in a database search HMM As a query entry , for example PFAM and SMART.
  • Markov chain ：
    • A data structure , By an initial state , A finite 、 A discrete set of possible states and a transition function that describes how to move from one state to the next . Also known as finite state machine .

      • 

  • One HMM Contains a set of defined states , It can be used to describe the position of each position in a sequence “ state ” Model of .

    • 

  • Hidden Markov model structure diagram
    • Icon ：

      

    • 1. The bottom box is a list of Main States , It may correspond to the residue of this amino acid sequence
    • 2. The second line consists of an insert state , This state contrast models the region in which two changes are made due to the necessary insertion
    • 3. The third line is composed of the deletion state corresponding to the vacancy represented by a circle . Provide a way to skim a column in a multiple sequence alignment . The emission generates the observation sequence in the comparison
    • State influencing factors
      • Transition probability and launch probability
        • The transition probability describes the path of the hidden state sequence along the Markov chain ;
        • The transmission probability is described as the matching state of the comparison time ;
  • HMMER Software ： Command line and web version
    • hmmbuild： A multiple sequence alignment can be read to create a spectrum HMM
    • hmmsearch： Appoint HMM And the database
      • Icon

        

    • hmmcalibrate： take 5000 Random sequence alignment to spectrum HMM On , Match the scores to a mechanism distribution , And calculate the parameters that are very necessary for estimating the statistical significance of database matching .
  • HMMER3 Improvement
    • Speed up
    • Provide HMMER Web version
    • Icon

      

•   Welcome to join the group , Or add VX：bbplayer2021, Invite in
•