Key points are not available for this paper at this time.
In the analysis of complex peptide mixtures by MS-based proteomics, many more peptides elute at any given time than can be identified and quantified by the mass spectrometer. This makes it desirable to optimally allocate peptide sequencing and narrow mass range quantification events. In computer science, intelligent agents are frequently used to make autonomous decisions in complex environments. Here we develop and describe a framework for intelligent data acquisition and real-time database searching and showcase selected examples. The intelligent agent is implemented in the MaxQuant computational proteomics environment, termed MaxQuant Real-Time. It analyzes data as it is acquired on the mass spectrometer, constructs isotope patterns and SILAC pair information as well as controls MS and tandem MS events based on real-time and prior MS data or external knowledge. Re-implementing a top10 method in the intelligent agent yields similar performance to the data dependent methods running on the mass spectrometer itself. We demonstrate the capabilities of MaxQuant Real-Time by creating a real-time search engine capable of identifying peptides “on-the-fly” within 30 ms, well within the time constraints of a shotgun fragmentation “topN” method. The agent can focus sequencing events onto peptides of specific interest, such as those originating from a specific gene ontology (GO) term, or peptides that are likely modified versions of already identified peptides. Finally, we demonstrate enhanced quantification of SILAC pairs whose ratios were poorly defined in survey spectra. MaxQuant Real-Time is flexible and can be applied to a large number of scenarios that would benefit from intelligent, directed data acquisition. Our framework should be especially useful for new instrument types, such as the quadrupole-Orbitrap, that are currently becoming available. In the analysis of complex peptide mixtures by MS-based proteomics, many more peptides elute at any given time than can be identified and quantified by the mass spectrometer. This makes it desirable to optimally allocate peptide sequencing and narrow mass range quantification events. In computer science, intelligent agents are frequently used to make autonomous decisions in complex environments. Here we develop and describe a framework for intelligent data acquisition and real-time database searching and showcase selected examples. The intelligent agent is implemented in the MaxQuant computational proteomics environment, termed MaxQuant Real-Time. It analyzes data as it is acquired on the mass spectrometer, constructs isotope patterns and SILAC pair information as well as controls MS and tandem MS events based on real-time and prior MS data or external knowledge. Re-implementing a top10 method in the intelligent agent yields similar performance to the data dependent methods running on the mass spectrometer itself. We demonstrate the capabilities of MaxQuant Real-Time by creating a real-time search engine capable of identifying peptides “on-the-fly” within 30 ms, well within the time constraints of a shotgun fragmentation “topN” method. The agent can focus sequencing events onto peptides of specific interest, such as those originating from a specific gene ontology (GO) term, or peptides that are likely modified versions of already identified peptides. Finally, we demonstrate enhanced quantification of SILAC pairs whose ratios were poorly defined in survey spectra. MaxQuant Real-Time is flexible and can be applied to a large number of scenarios that would benefit from intelligent, directed data acquisition. Our framework should be especially useful for new instrument types, such as the quadrupole-Orbitrap, that are currently becoming available. Mass spectrometry-based proteomics is generally performed in a shotgun format, where the proteome of interest is digested by a sequence specific protease and resulting peptides are analyzed by on-line liquid chromatography tandem mass spectrometry (LC MS/MS) 1The abbreviations used are:LC-MS/MSliquid chromatography-tandem mass spectrometryCIDcollision induced dissociationHCDhigher energy dissociationSILACstable isotope labeling with amino acids in cell cultureFDRfalse discovery rateLTQlinear trap quadrupoleMS/MStandem mass spectrometryOCXObject Linking and Embedding Control eXtensionSIMselected ion monitoring. 1The abbreviations used are:LC-MS/MSliquid chromatography-tandem mass spectrometryCIDcollision induced dissociationHCDhigher energy dissociationSILACstable isotope labeling with amino acids in cell cultureFDRfalse discovery rateLTQlinear trap quadrupoleMS/MStandem mass spectrometryOCXObject Linking and Embedding Control eXtensionSIMselected ion monitoring. (1Link A.J. Eng J. Schieltz D.M. Carmack E. Mize G.J. Morris D.R. Garvik B.M. Yates 3rd, J.R. Direct analysis of protein complexes using mass spectrometry.Nat. Biotechnol. 1999; 17: 676-682Crossref PubMed Scopus (2073) Google Scholar, 2Aebersold R. Mann M. Mass spectrometry-based proteomics.Nature. 2003; 422: 198-207Crossref PubMed Scopus (5596) Google Scholar, 3Yates 3rd, J.R. Gilchrist A. Howell K.E. Bergeron J.J. Proteomics of organelles and large cellular structures.Nat. Rev. Mol. Cell Biol. 2005; 6: 702-714Crossref PubMed Scopus (345) Google Scholar, 4Walther T.C. Mann M. Mass spectrometry-based proteomics in cell biology.J. Cell Biol. 2010; 190: 491-500Crossref PubMed Scopus (307) Google Scholar). Complex protein mixtures can contain thousands of proteins and an even much larger number of peptides are generated by the enzymatic digestion. As a result many peptides elute at a given time during chromatographic separation and the mass spectrometer needs to schedule peptide fragmentation events based on the peptide mass and intensity information in the MS spectra (“data dependent acquisition”). A widely used acquisition scheme is a topN method in which the mass spectrometer continuously cycles through full MS scans that are each followed by up to N precursor isolation and fragmentation events (MS/MS scans). In complex mixtures MS scans contain many more precursors than can be fragmented in the available time before the next full scan (5Michalski A. Cox J. Mann M. More than 100,000 Detectable Peptide Species Elute in Single Shotgun Proteomics Runs but the Majority is Inaccessible to Data-Dependent LC-MS/MS.J. Proteome Res. 2011; 10: 1785-1793Crossref PubMed Scopus (480) Google Scholar). To select precursors for fragmentation, the manufacturer’s software controlling the instrument sorts the precursor ions detected in each MS scan by intensity and also applies certain filters such as minimum signal intensity, charge state, and avoidance of already fragmented precursors. Inclusion and exclusion lists containing peptide masses of interest or those deemed uninteresting can also be used (6Rudomin E.L. Carr S.A. Jaffe J.D. Directed sample interrogation utilizing an accurate mass exclusion-based data-dependent acquisition strategy (AMEx).J. Proteome Res. 2009; 8: 3154-3160Crossref PubMed Scopus (32) Google Scholar, 7Schmidt A. Claassen M. Aebersold R. Directed mass spectrometry: towards hypothesis-driven proteomics.Curr. Opin. Chem. Biol. 2009; 13: 510-517Crossref PubMed Scopus (80) Google Scholar, 8Beck M. Claassen M. Aebersold R. Comprehensive proteomics.Curr. Opin. Biotechnol. 2011; 22: 3-8Crossref PubMed Scopus (72) Google Scholar). For maximizing identification success in shotgun proteomics, more elaborate selection schemes have been developed. For example, the Coon group has implemented a decision tree algorithm that schedules precursors for fragmentation either by collision-induced dissociation (CID) or by electron transfer dissociation based on their mass and charge (9Swaney D.L. McAlister G.C. Coon J.J. Decision tree-driven tandem mass spectrometry for shotgun proteomics.Nat. Methods. 2008; 5: 959-964Crossref PubMed Scopus (268) Google Scholar). liquid chromatography-tandem mass spectrometry collision induced dissociation higher energy dissociation stable isotope labeling with amino acids in cell culture false discovery rate linear trap quadrupole tandem mass spectrometry Object Linking and Embedding Control eXtension selected ion monitoring. liquid chromatography-tandem mass spectrometry collision induced dissociation higher energy dissociation stable isotope labeling with amino acids in cell culture false discovery rate linear trap quadrupole tandem mass spectrometry Object Linking and Embedding Control eXtension selected ion monitoring. In computer science “intelligent software agents” are constructed that make decisions in complex environments (10Russell, S. J., Norvig, P., (2009) Artificial Intelligence: A Modern Approach, 3rd Ed., Prentice hall, Englewood Cliffs, NJGoogle Scholar). Examples of intelligent agents are “spiders” that scour the web for search engines or software controlling mobile robots. Here we set out to conceptualize and construct an intelligent agent framework for proteomics and evaluate its applications to selected examples. The agent was implemented in the MaxQuant computational proteomics environment, which performs feature detection in raw MS data files, extracts high accuracy mass and quantification values, searches MS/MS data, and includes downstream bioinformatic analysis tools (11Cox J. Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification.Nat. Biotechnol. 2008; 26: 1367-1372Crossref PubMed Scopus (9214) Google Scholar, 12Cox J. Mann M. Computational principles of determining and improving mass precision and accuracy for proteome measurements in an Orbitrap.J. Am. Soc. Mass Spectrom. 2009; 20: 1477-1485Crossref PubMed Scopus (59) Google Scholar, 13Cox J. Neuhauser N. Michalski A. Scheltema R.A. Olsen J.V. Mann M. A Peptide the MaxQuant Proteome Res. 2011; 10: PubMed Scopus Google Scholar). isotope patterns and stable isotope labeling with amino in cell culture patterns based on information from MS scans and makes decisions data acquisition within the chromatographic time with the was by a software available by Object Linking and Embedding Control eXtension for Object Linking and Embedding Control engine for acquisition of tandem mass Chem. 2008; PubMed Scopus Google Scholar). We demonstrate the capabilities of MaxQuant Real-Time in applications to and in peptide based of the real-time peptide search engine as well as quantification of SILAC pairs whose ratios be from scans or The intelligent agent makes of the instrument which to the mass spectrometer. The agent much of for acquisition and real-time analysis It acquired information to analysis and with acquisition in a data-dependent The analysis of data the implemented from the MaxQuant computational proteomics for the of as well as the of isotope and SILAC have been to with the chromatographic data the full scan data is becoming available. For the of the a large number of are which are in an at the of the This makes the software to and the of the the is and the are to the instrument acquisition acquisition values, and based and with the liquid chromatography For each scan and are The of the and the mass with the The of the number of in and the data or For the and selected ion scan the is to MS and the time is For the scan the is to MS/MS with the isolation the or collision energy and the For the minimum is based on the as in the and the to the of the precursor For higher energy the minimum is set to and the as for the intelligent agent we also have the to out large of For example, the precursor mass information is much more accurate than that by to by the which makes the to the precursor in This is especially useful the isotope patterns of specific peptides of interest with were with and or with for in high modified of those amino acids and with extracts were by and of and J.D. by in a from Res. PubMed Scopus Google Scholar, M. S. R.A. Mann M. of to by of PubMed Scopus Google Scholar). The resulting extracts were in the by protein and to as J. M. Cox J. Mann M. isotope labeling by amino acids in cell culture and proteome of to a of 2008; PubMed Scopus Google Scholar). In extracts were to using for 30 at and were by for at in the was at protein and for at The was to by of and sequencing was to at was by of to and a to peptides was onto J. Mann M. and for and sample in Chem. 2003; PubMed Scopus Google for and to peptides were using and the was an high performance liquid chromatography of peptides were onto chromatography The were with tandem mass analysis was performed on a mass spectrometer at The was at to in to in to in to in at of at to in at and at and and The and MaxQuant Real-Time were set up with the For survey scans was set to and to with a ion time of and a scan range of to MS/MS scans were performed in the with a of and a ion time of scans were at a of with a of and a ion time of time to the was with ion trap collision and were set to ms, and and the isolation for precursor selection set to The mass range for scans was based on the G.C. in and applications of ion trap J. Mass Spectrom. Scopus Google Scholar). For the we used exclusion of with The resulting raw data were analyzed with the MaxQuant proteomics computational framework (11Cox J. Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification.Nat. Biotechnol. 2008; 26: 1367-1372Crossref PubMed Scopus (9214) Google Scholar). mass J. Mann M. mass by of peptide mass Am. Soc. Mass Spectrom. 2011; 22: PubMed Scopus Google was performed and the resulting accurate precursor masses to the database with the MaxQuant with a of J. Neuhauser N. Michalski A. Scheltema R.A. Olsen J.V. Mann M. A Peptide the MaxQuant Proteome Res. 2011; 10: PubMed Scopus Google was used to search the acquired spectra the fragmentation spectra of the database with a mass of was set as to and also to and a of of was set as and protein and as The false discovery rate was set to for peptides and proteins and the minimum peptide to amino analysis of the data by MaxQuant was performed in the and R. R. A for and 5: Scholar). The data used for analysis are at Here we set out to a and framework for intelligent data acquisition and real-time database searching for proteomics The intelligent agent was in as of the MaxQuant computational proteomics The framework is and to of we of the of the intelligent agent to of data acquisition in shotgun In the mass spectrometer is by high and software by the It the to acquisition methods in the high software on a which are the mass spectrometer prior to the where the for the software running on the mass spectrometer. In the of the of software is The software spectra in a data dependent and the acquired data to the high software on the also of the MS data as it is acquired and from raw data The intelligent agent is implemented using the and software of MaxQuant and on the as the software This is MaxQuant Real-Time and It with the acquisition software in the mass spectrometer through the high software it scan data as are in to of the mass spectrometer, MaxQuant Real-Time can its decisions on the full capabilities by the MaxQuant computational proteomics This can be on the real-time data as well as data from prior MS acquisition. can also be directed by from downstream bioinformatic the intelligent agent full scans and the resulting data from the instrument as it available to isotope and SILAC pair MaxQuant real-time isotope patterns or SILAC fragmentation scan events are on the As as the fragmentation data the is analyzed by the real-time search engine and its can be We the intelligent agent in the of and already implemented in MaxQuant MaxQuant Real-Time by with instrument feature detection from data, of MS data onto the as well as from bioinformatic is to the for acquisition and the A in real-time data analysis is the that information is available. In the of the isotope patterns have been at the time that the directed acquisition decision has to be We by the feature detection of MaxQuant (11Cox J. Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification.Nat. Biotechnol. 2008; 26: 1367-1372Crossref PubMed Scopus (9214) Google Scholar). For the MS the and detection can be used the full information is For and of isotope the are The applies to the detection of SILAC or stable isotope To of we the data from new spectra to the data of each and the isotope and SILAC pair detection a is detected or the be to an isotope within an time it is from the data The intelligent agent can also information from to the real-time This is a of the or exclusion In to charge and time we minimum and which are during the SILAC and information as well as intensity, peptide as well as for directed data acquisition decisions the of the intelligent The is at the and The agent of the mass spectrometer such as raw for scan as well as the to and it data acquisition. the the agent MS isotope patterns and SILAC precursor and MS/MS events. In of a the agent a scan is desirable on an isotope or SILAC pair and mass to from prior or from bioinformatic analysis is from in the MaxQuant computational proteomics that the intelligent agent with the data acquisition on the mass spectrometer, that is in the scan events of To data dependent with the intelligent we set out to a topN method capable of a complex peptide in This strategy can be as a for the full scan data analysis as the isotope patterns are used for any decision such as determining to be decisions at would result in identification in each survey the detected isotope patterns were by intensity each full scan to an of at precursor ions for As a signal MS/MS scans were at a given set to of the be in the This that peptides are detected in the are in intensity and are at the of the topN the time the MS/MS is performed the signal intensity is to have To the sequencing for a SILAC a fragmentation was for the more of the To performance we where an isotope is a of the detection from MaxQuant it is to the isotope of a peptide during its time and its the chromatography is on 30 As a fragmentation is which the of the MS/MS is for that isotope the isotope is an and its intensity has with the with the sequencing a new fragmentation is The at the intensity that fragmentation scans are that are likely to be than the based on the number of identified a of is in the of the chromatography exclusion fragmentation scans are which can be to of peptides at the with the number of scans an and a MaxQuant real-time directed acquisition. more scans for the of the by the intelligent agent during the of the scan This is in a time from to for a top10 an of scan in the the intelligent agent more a top10 with the resulting in similar identification the the time the can be we to that the intelligent agent than the method as it be to a higher scan and sequencing We constructed a of the search engine J. Neuhauser N. Michalski A. Scheltema R.A. Olsen J.V. Mann M. A Peptide the MaxQuant Proteome Res. 2011; 10: PubMed Scopus Google to real-time identification of the fragmentation spectra. The resulting is used to the real-time decisions of the intelligent but is during analysis with which can information for The was to the identification within the by the applied top10 fragmentation for a of time fragmentation at a of even than on a linear quadrupole instrument A. E. A. A. N. Cox J. Mann M. S. Mass spectrometry-based proteomics using a quadrupole mass To that time are we the a of on as a of on and on the and the SILAC and the database used by J. Neuhauser N. Michalski A. Scheltema R.A. Olsen J.V. Mann M. A Peptide the MaxQuant Proteome Res. 2011; 10: PubMed Scopus Google was by to of the data each of the mass peptide also the full sequence The search was performed a database with a of and with its We that more than of the peptides in the used SILAC The resulting database of peptides of which the of the algorithm were resulting in search of MaxQuant real-time than 30 in the proteome of This that more search are SILAC information for the precursor peptide is available number of or in the peptide amino from the database are The of a database in the real-time search for of a at a given false discovery As the identified is in real-time is to a for which the are in a with a to or peptide The of each identification is the and the for the selected We applied at an of and resulting in an of and that are used for real-time data decision and for the data As in real-time higher peptide ratios be desirable to false the has a which we to be to the peptides the of the whose identification be by the precursor mass to a larger To we the of the MS/MS spectra in the data which in a for the of generated data can result in high mass accuracy in the range J. Mann M. Computational principles of determining and improving mass precision and accuracy for proteome measurements in an Orbitrap.J. Am. Soc. Mass Spectrom. 2009; 20: 1477-1485Crossref PubMed Scopus (59) Google during acquisition mass can by to such as in This mass and exclusion Peptide identification is to by the mass of the mass spectrometer in We the mass of the real-time detected precursor mass from the in a of peptide peptide of and a mass based on the mass For the chromatography used the high peptide are generally within the during which time a search of is used and is applied to the search in MaxQuant J. Mann M. mass by of peptide mass Am. Soc. Mass Spectrom. 2011; 22: PubMed Scopus Google As as the is the precursor mass are continuously by the prior to database search with a search of To demonstrate the of the we a mass of an which was As a the real-time mass an from the of and exclusion have to be with large mass to mass during the which is by the real-time search engine the number of is to than by the from to in a This of mass the which to be in a in any is for the data analysis software has and has to data isotope patterns and the mass and time to be applied that result in mass precision in the range J. Mann M. mass by of peptide mass Am. Soc. Mass Spectrom. 2011; 22: PubMed Scopus Google Scholar). In to masses on the mass spectrometer, method is more especially for complex mixtures where the mass can be in feature full The of MaxQuant Real-Time to mass and time an during the data acquisition up a of data acquisition schemes by the intelligent We showcase the data acquisition for a protein of in the of a and the identification of peptides likely modified versions of already identified peptides. In the example, a MaxQuant Real-Time sample analysis was performed to a acquisition with fragmentation spectra acquisition for the topN peptide data analysis using the MaxQuant data analysis bioinformatic analysis in a of proteins with gene ontology (GO) the were selected and to the chromatographic through peptide and fragmentation spectra. The charge as well as time of chromatographic were used to an In a analysis of the sample MaxQuant Real-Time peptides for analysis of in MaxQuant to the identification of peptide up protein A of protein were to the term, peptide that were to the to MaxQuant Real-Time for was in of the and for for the of strategy is to any or for any of proteins of In shotgun proteomics modified peptides as pairs of peptides with a defined mass at In a of a directed analysis but time using the of a real-time we directed the intelligent agent to the of peptide pairs with a defined mass and time a Here we applied to peptides with a of isotope were for fragmentation in mass by a of an real-time identification of a peptide that more than We on SILAC pairs where the SILAC was to be the for the and modified the real-time data we identified the to be versions of the identified peptides as from their identification a which in be applied to any The the and the signal in a also termed a in mass for ions are to and a range the of identification and the precision of In scans a to by the trap with a mass Here we to for which the SILAC is the detection in survey by the precursor ions in MaxQuant Real-Time a isotope with a high intensity for fragmentation, a scan was the scan with a constructed This range is either based on identification the SILAC of the scan the is the or based on the detected charge state, the SILAC for the of to The is to a or higher mass selection of and on the and of the mass for the isolation in the linear trap of the mass spectrometer by of that strategy of a based on real-time information information in the data and to data analysis by on survey scan a isotope of a precursor ion at was identified In data the fragmentation identified the as the peptide with sequence and SILAC the SILAC peptide be from the MS survey spectra the its signal is the This scan the time with the full scan to range for the SILAC pair with a in the number of ions for the peptide of interest This is in the intensity a which are to for the The number of ions can be with the ion time in the for each which were and resulting in ions and ions at intensity for the scan and the scan the resulting extracts the SILAC from the and of the a of spectra were during a MaxQuant Real-Time analysis of a data by MaxQuant to the of of scan events with peptide peptide at data of the resulting peptide were with SILAC ratios and of were detected by MaxQuant Real-Time as and to a scan acquisition. The ratios from survey scan data to range The time for the scans was of the the strategy on mass is for complex the sample used is of high for time that higher quantification for selected precursors. Shotgun proteomics a large number of peptides to be fragmented and many more than can be even by mass (5Michalski A. Cox J. Mann M. More than 100,000 Detectable Peptide Species Elute in Single Shotgun Proteomics Runs but the Majority is Inaccessible to Data-Dependent LC-MS/MS.J. Proteome Res. 2011; 10: 1785-1793Crossref PubMed Scopus (480) Google Scholar). Here we have the of an intelligent which a framework to allocate to of interest in a agents are already widely used in many of information and that is also a in dependent decisions have been performed in MS/MS for many implemented at the in the instrument software for data acquisition. The intelligent agent and a
Graumann et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: