Apparatus for identifying peptides and proteins by mass spectrometry

Abstract

A method of identifying a protein, polypeptide or peptide by means of mass spectrometry and especially by tandem mass spectrometry is disclosed. The method preferably models the fragmentation of a peptide or protein in a tandem mass spectrometer to facilitate comparison with an experimentally determined spectrum. A fragmentation model is used which takes account of all possible fragmentation pathways which a particular sequence of amino acids may undergo. A peptide or protein may be identified by comparing an experimentally determined mass spectrum with spectra predicted using such a fragmentation model from a library of known peptides or proteins. Alternatively, a de novo method of determining the amino acid sequence of an unknown peptide using such a fragmentation model may be used.

1 . A method of identifying the most probable amino acid sequences which would account for the mass spectrum of a protein or peptide, said method comprising the steps of: 
 (a) producing a processable mass spectrum from said peptide; and    (b) using a fragmentation model to calculate the likelihood that any given trial amino acid sequence would account for said processable spectrum, said fragmentation model comprising the step of summing probabilistically a plurality of fragmentation routes which together represent the possible ways that said trial sequence might fragment in accordance with a set of predefined rules, each said fragmentation route being assigned a prior probability appropriate to the chemical processes involved.    
   
   
       2 . A method as claimed in  claim 1 , wherein said plurality of fragmentation routes represent all the possible ways that a said trial sequence might fragment.  
   
   
       3 . A method as claimed in  claim 2 , wherein the sum over all the possible fragmentation routes is calculated in polynomial time.  
   
   
       4 . A method as claimed in  claim 2  or  3 , wherein the sum over all the possible fragmentation routes is calculated in a time proportional to the number of amino acid residues in the peptide.  
   
   
       5 . A method as claimed in  claim 1 , wherein said fragmentation model is described using one or more Markov chains to represent one or more series of ions in which the probability of an ion being observed is influenced by whether or not an adjacent ion in the series was observed.  
   
   
       6 . A method as claimed in  claim 1 , wherein said prior probability takes into account the proton affinities of amino acid residues.  
   
   
       7 . A method as claimed in  claim 1 , wherein said prior probability takes into account the physical strength of the bonds broken in a proposed fragmentation route.  
   
   
       8 . A method as claimed in  claim 1 , wherein said fragmentation model includes the production of at least the b and y″ series of ions, wherein said b series is defined as comprising ions representing the N-terminal amino acid residue of the trial sequence and the loss of the C-terminal amino acid residues and said y″ series is defined as comprising ions representing the C-terminal amino acid residue and the loss of N-terminal amino acid residues.  
   
   
       9 . A method as claimed in  claim 8 , wherein said fragmentation model includes the production of the a series of ions, wherein said a series is defined as comprising b series ions which have lost CO.  
   
   
       10 . A method as claimed in  claim 8 , wherein said fragmentation model includes the production of z″ series ions, wherein said z″ series is defined as comprising y″ series ions which have lost NH 3 .  
   
   
       11 . A method as claimed in  claim 8 , wherein said fragmentation model includes the production of ions which have lost NH 3  and/or H 2 O.  
   
   
       12 . A method as claimed in  claim 8 , wherein said fragmentation model includes the production of immonium ions equivalent to the loss of CO and H from amino acid residues.  
   
   
       13 . A method as claimed in  claim 8 , wherein said fragmentation model includes the generation of sub-sequences of amino acids which begin and end at amino acid residues internal to the unknown peptide.  
   
   
       14 . A method as claimed in  claim 1 , wherein a said trial sequence is obtained from one or more libraries or databases containing sequences or partial sequences of known peptides and proteins.  
   
   
       15 . A method as claimed in  claim 1 , wherein a said trial sequence is generated pseudo-randomly using a de-novo sequencing method.  
   
   
       16 . A method as claimed in  claim 1 , wherein said fragmentation model is used to calculate the likelihood of amino acid sequences comprised in an existing protein or peptide database accounting for an experimentally observed mass spectrum of a peptide.  
   
   
       17 . A method as claimed in  claim 16 , wherein only sequences or partial sequences having a molecular weight within a given range are selected from said database for input into said fragmentation model.  
   
   
       18 . A method as claimed in  claim 1 , wherein a likelihood factor is assigned to each trial amino acid sequence considered.  
   
   
       19 . A method as claimed in  claim 1 , wherein the probability of a trial amino acid sequence accounting for said processable mass spectrum is calculated using Bayes' theorem wherein a prior probability assigned to a given trial amino acid sequence is multiplied by a likelihood factor which reflects the degree of agreement between a predicted mass spectrum resulting from said given trial amino acid sequence and said processable mass spectrum.  
   
   
       20 . A method as claimed in  claim 19 , wherein said prior probability includes the average natural abundances of amino acid residues.  
   
   
       21 . A method as claimed in  claim 19 , wherein said prior probability is influenced by the presumed genus or origin of said protein or peptide.  
   
   
       22 . A method as claimed in  claim 1 , wherein said processable mass spectrum comprises an observed mass spectrum.  
   
   
       23 . A method as claimed in  claim 1 , wherein said processable mass spectrum is obtained by converting multiply-charged ions and isotopic clusters of ions to a single intensity value at the mass-to-charge ratio corresponding to a singly-charged ion of the lowest mass isotope.  
   
   
       24 . A method as claimed in  claim 23 , further comprising the step of calculating an uncertainty value for the actual mass and the probability that a peak at that mass-to-charge ratio has actually been observed.  
   
   
       25 . A method as claimed in  claim 24 , wherein said uncertainty value is based on the standard deviation of a Gaussian peak representing the processed peak.  
   
   
       26 . A method as claimed in  claim 24 , wherein the probability that a peak is actually observed is based on the signal-to-noise ratio of the peak in the observed spectrum.  
   
   
       27 . Apparatus for identifying the most likely sequences of amino acids in an unknown peptide, said apparatus comprising a mass spectrometer for generating a mass spectrum of a said unknown peptide and data processing means programmed to: 
 (a) process data generated by said mass spectrometer to produce a processable mass spectrum; and    (b) calculate the likelihood that any given trial amino-acid sequence would account for said processable spectrum using a fragmentation model which sums probabilistically over a plurality of fragmentation routes which together represent the possible ways that said trial sequence might fragment in accordance with a set of predefined rules, each said fragmentation route being assigned a prior probability appropriate to the chemical processes involved.    
   
   
       28 . Apparatus as claimed in  claim 27 , wherein said mass spectrometer comprises a tandem mass spectrometer.  
   
   
       29 . Apparatus as claimed in  claim 27 , wherein said mass spectrometer further comprises a Time of Flight mass analyzer.  
   
   
       30 . Apparatus as claimed in  claim 27 , wherein said mass spectrometer further comprises an electrospray ionization source into which an unknown peptide sample may be introduced.  
   
   
       31 . A method of identifying a most probable amino acid sequence(s) which would account for a fragmentation mass spectrum of a protein or peptide, said method comprising the steps of: 
 producing the fragmentation mass spectrum from said protein or peptide;    providing a plurality of trial amino acid sequences;    using a fragmentation model to calculate a likelihood factor that any given trial amino acid sequence would account for said fragmentation mass spectrum, said fragmentation model comprising the step of summing probabilistically a plurality of fragmentation routes which together represent the possible ways that said trial amino acid sequence might fragment in accordance with a set of predefined rules, each said fragmentation route being assigned a prior probability appropriate to the chemical processes involved; and    selecting one or more of said trial amino acid sequences which have the highest likelihood factor(s) as being the most probable amino acid sequence(s) of said protein or peptide.