Center for Mass Spectrometry and Proteomics Phone (612) (612)

Transcription

1 Outline Database search types Peptide Mass Fingerprint (PMF) Precursor mass-based Sequence tag Results comparison across programs Manual inspection of results Terminology Mass tolerance MS/MS search FASTA Tandem MS Precursor mass Product ion Theoretical mass Experimental mass Sequence tag de novo sequencing

2 Three Strategies for Protein Inference from Peptide MS or MS/MS Data DATA INPUT: Peptide Mass List (MS1) Peptide MS/MS (MS2) Search TYPE: Peptide Mass Fingerprint (PMF) (historical) Precursor Mass-based 2 Sequencebased (de novo or mass tag) 1 3

3 1 Peptide Mass Fingerprint (PMF) Historical: after 1D or 2D in-gel Digestion of single bands or spots Current applications: limited; decreasing utility Component Input (Data) Target Search Parameters Description Peptide Mass List Mass type (Mr or MH+) Average or Monoisotopic Protein FASTA Sequence Database Proteolytic enzyme # Missed enzyme cleave sites Amino acid modifications Mass tolerance

4 Absolute Intensity 1 Peptide Mass Fingerprint: INPUT UNKNOWN PROTEIN (amino acid sequence): GLSDGEWQQVLNVWGKVEADIAGHGQEVLIRLFTG HPETLEKFDKFKHLKTEAEMKASEDLKKHGTVVLT ALGGILKKKGHHEAELKPLAQSHATKHKIPIKYLE FISDAIIHVLHSHPGNFGADAQGAMTKALELFRND IAAKYKELGFQG B) Acquire MALDI-TOF Mass Spectrum of peptide mixture A) In gel Trypsin digestion GLSDGEWQQVLNVWGK VEADIAGHGQEVLIR LFTGHPETLEK TEAEMK ASEDLK HGTVVLTALGGILK GHHEAELKPLAQSHATK YLEFISDAIIHVLHSHPGNFGADAQGAMTK ALELFR NDIAAK ELGFQG m/z C) Experimental Peptide Mass (or m/z) List m/z Peptide NDIAAK ELGFQG ASEDLK TEAEMK ALELFR LFTGHPETLEK HGTVVLTALGGILK VEADIAGHGQEVLIR GLSDGEWQQVLNVWGK GHHEAELKPLAQSHATK YLEFISDAIIHVLHSHPGNFGADAQGAMTK

5 1 PMF TARGET: Protein FASTA Sequence Database >sp P B_HUMAN protein beta/alpha OS=Homo sapiens GN=YWHAB PE=1 SV=3 MTMDKSELVQKAKLAEQAERYDDMAAAMKAVTEQGHELSNEERNLLSVAYKNVVGARRSS WRVISSIEQKTERNEKKQQMGKEYREKIEAELQDICNDVLELLDKYLIPNATQPESKVFY LKMKGDYFRYLSEVASGDNKQTTVSNSQQAYQEAFEISKKEMQPTHPIRLGLALNFSVFY YEILNSPEKACSLAKTAFDEAIAELDTLNEESYKDSTLIMQLLRDNLTLWTSENQGDEGD AGEGEN >sp P B_HUMAN Isoform Short of protein beta/alpha OS=Homo sapiens GN=YWHAB MDKSELVQKAKLAEQAERYDDMAAAMKAVTEQGHELSNEERNLLSVAYKNVVGARRSSWR VISSIEQKTERNEKKQQMGKEYREKIEAELQDICNDVLELLDKYLIPNATQPESKVFYLK MKGDYFRYLSEVASGDNKQTTVSNSQQAYQEAFEISKKEMQPTHPIRLGLALNFSVFYYE ILNSPEKACSLAKTAFDEAIAELDTLNEESYKDSTLIMQLLRDNLTLWTSENQGDEGDAG EGEN >sp P E_HUMAN protein epsilon OS=Homo sapiens GN=YWHAE PE=1 SV=1 MDDREDLVYQAKLAEQAERYDEMVESMKKVAGMDVELTVEERNLLSVAYKNVIGARRASW RIISSIEQKEENKGGEDKLKMIREYRQMVETELKLICCDILDVLDKHLIPAANTGESKVF YYKMKGDYHRYLAEFATGNDRKEAAENSLVAYKAASDIAMTELPPTHPIRLGLALNFSVF YYEILNSPDRACRLAKAAFDDAIAELDTLSEESYKDSTLIMQLLRDNLTLWTSDMQGDGE EQNKEALQDVEDENQ >sp Q F_HUMAN protein eta OS=Homo sapiens GN=YWHAH PE=1 SV=4 MGDREQLLQRARLAEQAERYDDMASAMKAVTELNEPLSNEDRNLLSVAYKNVVGARRSSW RVISSIEQKTMADGNEKKLEKVKAYREKIEKELETVCNDVLSLLDKFLIKNCNDFQYESK VFYLKMKGDYYRYLAEVASGEKKNSVVEASEAAYKEAFEISKEQMQPTHPIRLGLALNFS VFYYEIQNAPEQACLLAKQAFDDAIAELDTLNEDSYKDSTLIMQLLRDNLTLWTSDQQDE EAGEGN >sp P G_HUMAN protein gamma OS=Homo sapiens GN=YWHAG PE=1 SV=2 MVDREQLVQKARLAEQAERYDDMAAAMKNVTELNEPLSNEERNLLSVAYKNVVGARRSSW RVISSIEQKTSADGNEKKIEMVRAYREKIEKELEAVCQDVLSLLDNYLIKNCSETQYESK VFYLKMKGDYYRYLAEVATGEKRATVVESSEKAYSEAHEISKEHMQPTHPIRLGLALNYS VFYYEIQNAPEQACHLAKTAFDDAIAELDTLNEDSYKDSTLIMQLLRDNLTLWTSDQQDD DGGEGNN >sp P S_HUMAN protein sigma OS=Homo sapiens GN=SFN PE=1 SV=1 MERASLIQKAKLAEQAERYEDMAAFMKGAVEKGEELSCEERNLLSVAYKNVVGGQRAAWR VLSSIEQKSNEEGSEEKGPEVREYREKVETELQGVCDTVLGLLDSHLIKEAGDAESRVFY LKMKGDYYRYLAEVATGDDKKRIIDSARSAYQEAMDISKKEMPPTNPIRLGLALNFSVFH YEIANSPEEAISLAKTTFDEAMADLHTLSEDSYKDSTLIMQLLRDNLTLWTADNAGEEGG EAPQEPQS etc UniProt (example) The Universal Protein Resource (UniProt) is a comprehensive resource for protein sequence and annotation data. UniProt is a collaboration between the European Bioinformatics Institute (EBI), the Swiss Institute of Bioinformatics (SIB) and the Protein Information Resource (PIR). Across the three institutes close to 150 people are involved through different tasks such as database curation, software development and support. Example: 20,292 entries for taxonomy: "Homo sapiens (Human) [9606] AND reviewed:yes

6 1 PMF Search Parameters Review: Trypsin Proteolytic Enzyme and 0 Missed Cleave Site (#MC) modifications Example: Theoretical Trypsin Digest Peptide Mass List (truncated) for a Single Protein: Trypsin = Enzyme Missed Cleave Sites = 0

7 1 PMF Search Parameters: Trypsin Proteolytic Enzyme and 1 Missed Cleave Site MSO = Methionine Oxidation (amino acid modification) artif modifications modifications

8 1 PMF Parameter: Peptide Mass Deviation (+/- m/z) High resolution measurement / Range: Mouse proteins with Trypsin peptide in this range Low resolution measurement 631 +/- 1 Range: Mouse proteins with Trypsin peptide in this range m/z

9 1 Mass Measurement Error Calculation Error Expression Fractional Error Expression Multiply by: ppm / Theoretical value 1,000,000 % / Theoretical value 100 where Δ = Experimental (or observed) Theoretical m/z value Example: Experimental/observed value (i.e., the data acquired by the mass spectrometer) = m/z Theoretical value (calculated from periodic table, after the peak is identified) = m/z Delta (Δ) = Error Expression Error Equation Error ppm (0.079/ )*1,000, ppm % (0.079/ )* % A USEFUL REFERENCE: Brenton AG, Godfrey AR. Accurate mass measurement: terminology and treatment of data. J Am Soc Mass Spectrom Nov;21(11):

10 1 MASCOT Peptide Mass Fingerprint Search

11 1 MASCOT PMF Search Parameter Page

12 1 PMF Search QUERY (Experimental Data) m/z Peptide NDIAAK ELGFQG ASEDLK TEAEMK ALELFR LFTGHPETLEK HGTVVLTALGGILK VEADIAGHGQEVLIR GLSDGEWQQVLNVWGK GHHEAELKPLAQSHATK YLEFISDAIIHVLHSHPGNFGADAQGAMTK Search Score Compare QUERY to THEORETICAL PEPTIDE MASS LIST for each protein in the database Parameters: Enzyme Missed cleave site Amino acid mods Mass tolerance Probability-based MOWSE Score (often, the protein with highest number of peptide matches has the highest score)

13 1

14 1

15 1 PMF Search: Results Interpretation Is the protein ID experimentally rational? Does the MW of protein in search results match MW determined by SDS-PAGE? Does the pi of protein in search results match pi determined by 2D-PAGE? Perform MS/MS if no protein match

16 Three Strategies for Protein Inference from Peptide MS or MS/MS Data DATA INPUT: Peptide Mass List (MS1) Peptide MS/MS (MS2) Search TYPE: Peptide Mass Fingerprint (PMF) (historical) Precursor Mass-based 2 Sequencebased (de novo or mass tag) 1 3

17 2 Tandem MS: Precursor Mass-based Search Component Input (Data) Target Search Parameters Description Precursor Mass & Product Ion Masses Charge state Average or Monoisotopic Protein FASTA Sequence Database Proteolytic enzyme # Missed enzyme cleave sites Amino acid modifications Mass tolerances: Peptide/Precursor Mass Product Ions Masses

18 2 Tandem MS: Precursor Mass-based Search 1. Each MS/MS spectra has 2 data-rich Components 2. Database search for each MS/MS spectra has 2 STEPS

19 2 Tandem MS: Precursor Mass-based Search REVIEW: Each MS/MS Spectra has Two Data-Rich Components In te n s ity, c o u n ts Example: m/z = Precursor m/z +TOF MS: Experiment 1, min from bruce w 2-5 humjak3 tryp.wiff a= e-004, t0= e COMPONENT 1: Intact Peptide Max counts m/z, amu In te n s ity, c o u n ts +TOF Product (612.8): Experiment 2, min from bruce w 2-5 humjak3 tryp.wiff a= e-004, t0= e a COMPONENT 2: Product Ion Mass (m/z) values and Intensities after Peptide Fragmentation a2b2 y b y y a3-nh3(2+) y y4 y7(2+) a4 y8(2+) b4 a5 y y y Max counts m/z, amu y9

20 Tandem MS: Precursor Mass-based Search STEP 1: Find Theoretical Peptides (from in silico Protein Digest) within user specified Precursor Mass Tolerance from MS1 spectrum Mass [M + H] +1 Protein GenBank ID# Peptide gi SRNLGPSTSR gi RASVDIGISR gi MoxRTFMoxISR gi QTYENYTR gi REEMHESR gi IDLVSMoxHSR gi IFLPGHYAR gi FMLPGDTHR gi NMoxRQHDTR gi EVPETKDTR PEPTIDE CANDIDATES 2 evaluate 1 st candidate (next slide) etc etc etc

21 2 Tandem MS: Precursor Mass-based Search STEP 2: Generate Theoretical Product Ion Table for Peptide Candidate SRNLGPSTSR Theoretical Product Ion Table evaluate 1 st candidate Residue a b b-h2o b-nh3 y y-h2o y-nh3 S, Ser R, Arg N, Asn L, Leu G, Gly P, Pro S, Ser T, Thr S, Ser R, Arg

22 2 Tandem MS: Precursor Mass-based Search Compare theoretical product ions to experimental product Ions from MS2 spectrum

23 Intensity 2 REVIEW: b-type fragment ions NH3+ A E P T I R COOH NH2 NH2 A E P T A E P NH2 A E NH2 A m/z

24 Intensity 2 REVIEW: y-type fragment ions NH3+ A E P T I R COOH HOOC HOOC R I T P R I T HOOC R I HOOC R m/z

25 Intensity 2 REVIEW: b- and y-type Fragment ions: PEPTIDE TANDEM MASS SPECTRUM NH3+ A E P T I R COOH b y b y b y b y b y m/z

26 2 Tandem MS: Precursor Mass-based Search SUMMARY- for Each MS/MS Spectrum: Generate theoretical product ion tables for all peptide candidates Compare theoretical product ions to experimental product Ions from MS2 spectrum Score all candidate peptides Rank peptides by Score

27 2 Tandem MS: Precursor Mass-based Search STEP 1 Find Theoretical Peptide Matches from in silico Protein Digest in the range: Experimental Precursor Mass +/- Mass Tolerance STEP 2 For each candidate peptide from Step 1: Compare Theoretical Product Ions (b & y, etc) to Experimental Product ions (data) STEP 3 SCORE: many software programs/ algorithms Peptide rank PROTEIN Report and Grouping: many variations

28 2 Tandem MS: Precursor Massbased search Software at UM Sequest X!Tandem MASCOT

29 Three Strategies for Protein Inference from Peptide MS or MS/MS Data DATA INPUT: Peptide Mass List (MS1) Peptide MS/MS (MS2) Search TYPE: Peptide Mass Fingerprint (PMF) (historical) Precursor Mass-based 2 Sequencebased (de novo or mass tag) 1 3

30 3 Sequence-based Searches: WHEN? Unsequenced genome (protein(s) of interest are not in the database) HOMOLOGY-based search Search for amino acid MUTATIONS Search for LARGE NUMBERS of Post Translational Modifications (PTM s)

31 3 Tandem MS: Sequence-based Search STEP 1: Obtain [full or partial] amino acid sequence string directly from the spectrum by de novo Sequencing de novo (Latin) "from the beginning"

32 3 Tandem MS: Sequence-based Search STEP 1: de novo sequencing: amino acid sequence is determined from delta mass values for a series of successive peptide b- or y-type product ions from a Product Ion (MS/MS) spectrum EXAMPLE de novo sequencing: L 113 V 99 A 71 D 115 L 113 E 129 delta masses between peaks correspond to amino acid residue masses See amino acid residue masses:

33 3 Tandem MS: Sequence-based Search STEP 2: QUERY THE PROTEINS Example: AB Sciex Paragon Algorithm For each MS/MS spectrum, multiple, short sequences (taglets) are generated, scored, and matched against protein sequences. Sequence Temperature Values (STV) rank candidate amino acid sequence segments. STV is related to the number of taglet hits per region. Slide content: courtesy of Sean L. Seymour, AB Sciex

34 3 Tandem MS: Sequence-based Search STEP 3: Candidate peptide sequences within protein sequences are identified with Precursor mass AND sequence tag information Include: Enzyme specificity Mass tolerance Amino acid modifications Amino Acid mutations STEP 4: Candidates are Ranked and SCORED

35 3 Tandem MS: Sequence-based Search STEP 3b: Error-tolerant Sequence Tag Reconciliation Includes Amino Acid Mutation Search Component Combination: sequence tag & mass match Slide content: adapted from David L. Tabb, Vanderbilt University: Bioinformatics 2009, Baltimore MD

36 3 MS/MS Sequence-based Software at UM ProteinPilot (AB Sciex) PEAKS (BSI) Direct Tag (D Tabb Lab) MASCOT (Matrix Science LTD) Error Tolerant MS/MS Sequence Tag

37 3 MS/MS Sequence-based Software PepNovo (open source de novo) Frank et al, Anal Chem (2005) 77: PEAKS Studio (Commercial de novo) Ma et al, RCMS (2003) 17: 2337:2342 Lutefisk (early de novo tool) Taylor et al, RCMS (1997) 11: Direct Tag (open source tag inference) Tabb et al, J Proteome Res (2008) 7: InsPecT (open source tag identifier) Tanner et al, Anal Chem (2005) 77:

38 Maximize Identifications with Multiple Software Programs Single Dataset analyzed with multiple software programs: Blue = number of peptide matches Red = number of protein matches

39 Maximize Identifications with Multiple Software Programs

40 Search Algorithm Conclusions 1) Several different search algorithm types exist 2) Each program provides useful results - no unified method exists 3) The same data searched using different algorithms may yield different results (scoring and ranking schemes are distinct)

41 Manual Inspection of Individual MS/MS Spectra Journal requirement (in special cases; see Inspect SINGLE PEPTIDE HITS Site Localization of Post Translational Modifications (PTM s) Detection of unexpected gene products (alternative splice isoforms)

42 STEPS for Manual Inspection of MS/MS Spectra 1) Open RAW data; print spectrum (or, multiple zoomed-in regions of the spectrum) 2) Obtain list of Theoretical Fragment Ions (e.g., open MS Product tool 3) Label printout(s) of spectra with fragment ion types 4) DECISION: a. GOOD MATCH: all/most of the peaks (in the raw data) labeled with fragment ion types b. POOR MATCH: unassigned peaks in the raw data

43 Manual Inspection of MS/MS Spectra Example: RAW data labeled with Fragment Ion Names/Types IT4FYEQFSK IT4 (Fragment Ion Nomenclature Reference: P. Roepstorff & J. Fohlman (2005) Proposal for a Common Nomenclature for Sequence Ions in Mass Spectra of Peptides, Biomedical Spectrometry, 11(11), p 601