PeakInvestigator® Information
PeakInvestigator Introductory Seminar
- PeakInvestigator Introductory Overview
- Lipid Biomarkers Blind Discovery presentation
- Increased Quantity & Quality of Proteomics Identifications presentation
- Maximize Discovery of Robust Metabolite Features in Complex Sample Analysis with Dynamic Thresholding & Precision Centroiding presentation
- Mass Spectral Data Processing & IVA Examples presentation
PeakInvestigator Supplemental Materials
- Mass Spectral Characterization Files for PeakInvestigator document
- PeakInvestigator v2.0 statistical centroiding presentation
- Narrated 15-minute movie of the PeakInvestigator v1.0 presentation.
- PeakInvestigator v1.0 presentation slides (with narration)
- PeakInvestigator v1.0 presentation slides (without narration)
- PeakInvestigator v1.0 brochure
PeakInvestigator Software and MZmine PeakInvestigator™ Edition Demonstration
PeakInvestigator Frequently Asked Questions (FAQ)
Posters
2018
-
“Optimizing Information Content in Collision-Induced Fragmentation Spectra of Peptides for Better Sequencing" (ASMS)
Abstract:
The most common method of peptide fragmentation in mass spectrometry for sequencing is collision-induced dissociation (CID). However, peptide sequencing using CID data still remains challenging. We have previously shown that this is partly due to the limited information present in the spectra generated under sub-optimal fragmentation conditions. Further, current peptide sequencing algorithms utilize mostly a-, b- & y- ions along with their neutral losses which may lower the scores for correct sequence assignments because of other confounding fragment ions. In this study, we optimize the collision energy profile in silico to improve sequencing ion coverage from CID spectra generated from multiple collision energies.
Citation:
Sokkalingam N, Schneider L, Wright W, Ashrafi S, Tenderholt A, Peterson J, Duncan M. Optimizing Information Content in Collision-Induced Fragmentation Spectra of Peptides for Better Sequencing. Poster presented at: 66th Annual Am. Soc. Mass Spec.; 2018 June 3-7; San Diego, CA.
2017
-
“Isotopic Vector Angle Analysis to Identify Protein Isoforms in Mass Spectrometry" (HUPO)
Abstract:
There are over 1200 hemoglobin variants, most of which are SNPs (single nucleotide polymorphisms) that manifest in a single amino acid substitution, such as HbS that causes Sickle Cell trait when present as Hb AS and sickle cell disease when present as Hb SS. Current techniques used to detect these variants, such as HPLC and CE are limited in ability to identify the single amino acid substitution. Mass spectrometry (LC-MS) is a viable alternative if the normal and variant hemoglobin can be separated during the reverse phase chromatography; however, if they can not be separated and the mass shift between normal and variant is < 6 Da, then LC-MS alone may not be adequate to determine the mass of the variant. In this poster, we present a new method to statistically identify and quantify hemoglobin variants that form nearly complete mass overlaps in the mass spectrum based on the vector representation of their isotopic pattern and the Isotopic Vector Angle (IVA).
Citation:
Sokkalingam N, Schneider L, Herold D, Yang J. Isotopic Vector Angle Analysis to Identify Protein Isoforms in Mass Spectrometry. Poster presented at: 16th Int’l HUPO World Congress; 2017 Sep 17-20; Dublin, Ireland. -
“PeakInvestigator® Maximizes Discovery of Robust Metabolite Features in Complex Sample Analysis with Dynamic Thresholding & Precision Centroiding” (ASMS) (Metabolomics)
Abstract:
PeakInvestigator® (PI) is a new advanced signal processing method for statistically-driven centroiding of mass spectrometric data, which is fully automated. In this study, we compared PeakInvestigator to Standard Centroiding (MZmine Exact Mass) by analyzing three lipodomic samples (food plate homogenates from different diets) processed by LC/MS (Metabolomics 2015 Challenge). Metabolomics analyses were performed independently on the centroided masslists generated by the different methods to identify statistically-significant sample (plate)-discriminating features. The additional chromatographic features discovered with PeakInvestigator were due to a combination of locally-adaptive dynamic thresholding, statistical signal to noise discrimination, and better centroiding precision.
Citation:
Peterson J, Sokkalingam N, Wright W, Schneider L, Tenderholt A. PeakInvestigator® Maximizes Discovery of Robust Metabolite Features in Complex Sample Analysis with Dynamic Thresholding & Precision Centroiding. Poster presented at 65th Annual Am. Soc. Mass Spec.; 2017 June 4-8; Indianapolis, IN. -
“Centroiding Error and Isotopic Vectors for Improved LC/MS Analyses” (USHUPO)
Abstract:
PeakInvestigator® is a new fully-automated centroiding software that provides both mass and abundance confidence intervals for every MS peak called. This unique feature enables new approaches for downstream data analysis in LC/MS experiments. In this work we show how isotopic abundance precision can be used for determining the start and end times for chromatographic peaks and how this information translates into better kinetic information in stable isotope labeling experiments.
Citation:
Schneider L, Sokkalingam N, Price J, Hannemann A. Centroiding Error and Isotopic Vectors for Improved LC/MS Analyses. Poster presented at: 13th Annual US HUPO Conference; 2017 Mar 19-22; San Diego, CA, USA. -
“Centroiding with Statistical Confidence: Mass and Abundance Error Bars from PeakInvestigator™ 2.0” (MSACL)
Abstract:
Building chromatograms from LC/MS data typically requires the assignment of a single best estimate of the proper mass tolerance for peak alignment (Figure 1). This can result in incomplete or inaccurate chromatograms and peaklists. PeakInvestigator™ 2.0 provides statistically-valid mass confidence intervals for every peak call. The open-source LC/MS software package (MZmine-PI edition) was modified to use this information in building chromatograms (Figure 2). This new MS analysis capability allows each peak in an LC/MS experiment to be aligned with its own dynamic mass tolerance, adjusting to any statistical confidence level desired. Standard centroiding (Exact Mass in MZmine) produces just one chromatographic peak, which is resolved into two peaks only by using the Chromatogram Deconvol-ution feature of MZmine, but not by mass. In the same region, PeakInvestigator™ masslists are used by MZmine to resolve three distinct chromatographic peaks by their 98% mass confidence limits alone.
Citation:
Schneider L, Sokkalingam N, Tenderholt A. Centroiding with Statistical Confidence: Mass and Abundance Error Bars from PeakInvestigator™ 2.0. Poster presented at: MSACL 2017 US; 2017 Jan 22-26; Palm Springs, CA, USA.
2016
-
“Improved MS/MS Peptide Identification Using Advanced Peak Detection and Deconvolution” (HUPO)
Abstract:
Correct peptide identification depends upon several factors, including accurate masses obtained from tandem MS experiments and the ability to distinguish signal from noise. This is especially important in shotgun proteomics experiments where the complexity of the sample mixture might contain peptides with nearly-isobaric fragments that can overlap to create misidentified peaks, especially with low resolution mass analyzers. In this poster, we compare the peptide sequencing results from Proteome Discoverer (database search methodology) for masslists created by three different centroiding software packages: PeakInvestigator™, Apex Picking (Proteome Discoverer v1.4, Thermo Scientific) and OpenMS (PeakPickerHiRes 2.0.0). PeakInvestigator uses proprietary advanced signal processing methods for peak detection and deconvolution in a software-as-a-service platform. OpenMS utilizes standard, public domain, centroiding methods and the manufacturer’s software utilizes apex picking algorithms. Each procedure was verified against an LC/MS run of 216 synthetically-prepared peptides. The same LC/MS and analysis procedures were then applied to both a tryptic digest of a yeast proteome (S. cerevisiae) and a human phospho-proteome sample. PeakInvestigator masslists consistently provided higher Mascot peptide scores (Proteome Discoverer Ion Scores) than the other centroiding software packages on all three samples. PeakInvestigator increased the number of peptide hits by 29% (vs. Apex Picking) and 15% (vs. OpenMS) on the more complex phosphorylated human sample. PeakInvestigator also yielded 7% more phosphorylation sites versus OpenMS and 17% more than Apex Picking on this sample. Comparable results in both peptide numbers and unique peptides identified were obtained from all the centroiding packages on the less complex samples.
Citation:
Sidoli S, Kulej K, Garcia B, Sokkalingam N, Tenderholt A, Schneider L, Peterson J. Improved MS/MS Peptide Identification Using Advanced Peak Detection and Deconvolution. Poster presented at: 15th Int’l HUPO World Congress; 2016 Sep 18-21; Taipei, Taiwan. -
“How Thresholding Affects the Reliability of Peak Identifications” (Metabolomics)
Abstract:
The first stage in processing a raw (profile) mass spectrum typically involves the steps of baseline correction, peak detection (centroiding), and signal-to-noise threshold determination. Historically, all three of these techniques require the mass spectrometrist to assign multiple user-adjustable parameters based upon their judgment and experience. In this study we use receiver operating characteristic (ROC) analysis to compare and contrast various baselining and thresholding methods applied to a single, well-curated, time-of-flight (TOF) spectrum selected at random from a lipidomics LC/MS study. In the first part of this study we contrast results of a variety of abundance-driven thresholding methods applied to previously-centroided data, including the parameter-free method incorporated in the PeakInvestigator™ software. In the second part of this study we examine thresholding the raw spectrum prior to both standard centroiding and PeakInvestigator™ peak determination. The lack of significant differences between the thresholding methods strongly suggests that PeakInvestigator™ requiring no user-adjustable parameters, or parameters that can be derived directly from the data (without user intervention), will give the most consistent results. Furthermore, we show that automated thresholding of the profile spectrum before centroiding provides better true positive peak detection (as measured by increasing PPV), independent of the peak detection (centroiding) method.
Citation:
Sokkalingam N, Schneider L, Tenderholt A, Ashrafi S. How thresholding affects the reliability of peak identifications. Poster presented at: 12th Annual International Conference of the Metabolomics Society; 2016 Jun 27-30; Dublin, Ireland. -
“Deconvolution and Isotopic Vector Analysis for Improved Peak Identification” (ASMS)
Abstract:
The National High Magnetic Field Laboratory (NHMFL) continues to push the FT-ICR technique to its ultimate limits for mass resolution, mass range, and sensitivity. FT-ICR allows us to identify more than 100,000 components without prior separation, from which we correlate and ultimately predict the origin properties and composition, of complex mixtures such as crude oil and its distillates, vegetable oils, and wines. However, even with the highest field FT-ICR instruments, some compositions still elude complete characterization. As we show herein, PeakInvestigator™ allows us to extend the resolution of even FT-ICR mass analyzers and provides more accurate mass determination. We also introduce isotopic vector analysis as a way to qualify peak identifications and quantify the relative abundances of isotopically-tagged species.
Citation:
Sokkalingam N, Schneider L, Tenderholt A, Chu F, Corillo YE, Marshall AG. Deconvolution and isotopic vector analysis for improved peak identification. Poster presented at: 64th Annual Am. Soc. Mass Spec.; 2016 June 5-9; San Antonio, TX.
2015
-
“More Biomarkers Discovered with Spectral Deconvolution Centroiding” (HUPO)
Abstract Part 1:
Peak overlaps created by isotopomers, multiply-charged species, and intermingled isotopic patterns continue to confound mass spectral biomarker analysis. This problem presents itself in all mass analyzers, at any resolution. In this initial study with the Fiehn lab (UC Davis), we compared a new advanced signal processing method (PeakInvestigator™) to conventional spectral centroiding practice, examining enhanced biomarker discovery potential on two LC/MS lipodomic samples.The advanced signal processing in PeakInvestigator resulted in 41 (4.2%) more chromatographic features revealed from these LC/MS data, with 36 being true positives (88% Positive Predictive Value). After deisotoping, the results totaled to 23 monoisotopic true positive potential biomarker discoveries.
Abstract Part 2:
Peak overlaps created by isotopomers, multiply-charged species, and intermingled isotopic patterns continue to confound mass spectral biomarker analysis. This problem presents itself in all mass analyzers, at any resolution. In this followup study with the Fiehn lab (UC Davis), we compared a new advanced signal processing method (PeakInvestigator™) to conventional spectral centroiding practice, examining unique biomarker discovery potential across 93 patient samples from the SATURN Coronary Atheroma study. PeakInvestigator uniquely detected and deconvolved 6 new, real, statistically-robust chromatographic peaks from the LC/MS data set, which standard centroiding failed to detect in any sample. These data also showed that the 9 features uniquely reported by standard centroiding were uniformly low-abundance features statistically undifferentiated from noise.
Citation:
Schneider L, Sokkalingam N, Wright W, Tenderholt A, Lewitt M, Gerner L, Schmidt D, Cajka T, Fiehn O. More biomarkers discovered with spectral deconvolution centroiding. Poster presented at: 14th Int’l HUPO World Congress; 2015 Sep 27-30; Vancouver, BC, Canada.
[Download PDF of poster Part 1]
[Download PDF of poster Part 2] -
“Use of Spectral Deconvolution Centroiding to find More Robust, Sample-Discriminating Biomarkers” (HUPO)
Abstract:
Biomarker discovery via LC/MS is often limited byinability to detect and resolve overlapped nearly isobaric species that were also unresolved chromatographically. This study evaluated the ability of mass spectral post-processing with advanced signal processing for peak detection, deconvolution and centroiding (PeakInvestigator™) to reveal additional robust, sample-differentiating, biomarkers.
Citation:
Sokkalingam N, Schneider L, Wright W, Tenderholt A, Lewitt M, Gerner L, Schmidt D. Use of spectral deconvolution centroiding to find more robust, sample-discriminating biomarkers. Poster presented at: 14th Int’l HUPO World Congress; 2015 Sep 27-30; Vancouver, BC, Canada. -
“Data Acquisition and Processing to Apply Inter-Peak Isotope Spacing for Turnover Analysis” (HUPO)
Abstract:
Metabolic labeling with deuterium has been used to monitor turnover rates for proteins, lipids, DNA, and RNA in humans as well as model systems. Turnover is measured as time-dependent deuterium incorporation into newly synthesized molecules, using mass spectrometry to monitor changes in isotope distribution. The increase in deuterium changes the relative ratio of isotope abundances and a within-pattern isotopic mass spacing that is different than a simple increase of a neutron mass (mass defect). Simulations of the isotope pattern suggest that the changes in inter-peak distances due to changing ratios of mass defect incorporation into each isotope position (i.e. increasing 2H vs. 13C) should provide turnover information similar to changes in isotope abundance. Comparison against the theoretical isotope pattern provided a gold standard to evaluate whether each permutation increased or decreased accuracy. Using the optimized workflow, kinetic proteomics studies can gain additional confidence by leveraging inter-peak distance information.
Citation:
Naylor B, Porter M, Wilson E, Herring A, Price J. Data acquisition and processing to apply inter-peak isotope spacing for turnover analysis. Poster presented at: 14th Int’l HUPO World Congress; 2015 Sep 27-30; Vancouver, BC, Canada. -
“Discover More Biomarkers with PeakInvestigator™” (Metabolomics)
Abstract Part 1:
Peak overlaps created by isotopomers, multiply-charged species, and intermingled isotopic patterns continue to confound mass spectral biomarker analysis. This problem presents itself in all mass analyzers, at any resolution. In this initial study with the Fiehn lab (UC Davis), we compared a new advanced signal processing method (PeakInvestigator™) to conventional spectral centroiding practice, examining enhanced biomarker discovery potential on two LC/MS lipodomic samples.The advanced signal processing in PeakInvestigator resulted in 41 (4.2%) more chromatographic features revealed from these LC/MS data, with 36 being true positives (88% Positive Predictive Value). After deisotoping, the results totaled to 23 monoisotopic true positive potential biomarker discoveries.
Abstract Part 2:
Peak overlaps created by isotopomers, multiply-charged species, and intermingled isotopic patterns continue to confound mass spectral biomarker analysis. This problem presents itself in all mass analyzers, at any resolution. In this followup study with the Fiehn lab (UC Davis), we compared a new advanced signal processing method (PeakInvestigator™) to conventional spectral centroiding practice, examining unique biomarker discovery potential across 93 patient samples from the SATURN Coronary Atheroma study. PeakInvestigator uniquely detected and deconvolved 6 new, real, statistically-robust chromatographic peaks from the LC/MS data set, which standard centroiding failed to detect in any sample. These data also showed that the 9 features uniquely reported by standard centroiding were uniformly low-abundance features statistically undifferentiated from noise.
Citation:
Schneider L, Sokkalingam N, Wright W, Tenderholt A, Lewitt M, Gerner L, Schmidt D, Cajka T, Fiehn O. Discover more biomarkers with PeakInvestigator™. Poster presented at: 11th Annual International Conference of the Metabolomics Society; 2015 Jun 29-Jul 2; San Francisco, CA.
[Download PDF of poster Part 1]
[Download PDF of poster Part 2] -
“Discover More Biomarkers with PeakInvestigator™” (ASMS)
Abstract:
Peak overlaps created by isotopomers, multiply-charged species, and intermingled isotopic patterns continue to confound mass spectral biomarker analysis. This problem presents itself in all mass analyzers, at any resolution. In this followup study with the Fiehn lab (UC Davis), we compared a new advanced signal processing method (PeakInvestigator™) to conventional spectral centroiding practice, examining unique biomarker discovery potential across 93 patient samples from the SATURN Coronary Atheroma study. PeakInvestigator uniquely detected and deconvolved 6 new, real, statistically-robust chromatographic peaks from the LC/MS data set, which standard centroiding failed to detect in any sample. These data also showed that the 9 features uniquely reported by standard centroiding were uniformly low-abundance features statistically undifferentiated from noise.
Citation:
Schneider L, Sokkalingam N, Wright W, Tenderholt A, Cajka T, Fiehn O, Contrepois K, Mahmoudi S, Porter M, Naylor B, Herring A, Price J. Discover more biomarkers with PeakInvestigator™. Poster presented at: 63rd Annual Am. Soc. Mass Spec.; 2015 June 5-9; St. Louis, MO.
2014
-
“PeakInvestigator™: Software to Improve the Precision, Resolution, and Sensitivity of your Mass Analyzer” (ASMS)
Abstract:
Mass spectrometry is a critical tool for the identification of molecular ions. However, it can only measure the apparent mass and abundance of these ions. While mass accuracy depends on user calibration, the precision with which the mass and abundance can be determined depends on the software used to centroid and discriminate the peaks from confounding chemical and instrumental noise. Any software tool that improves the precision of the mass and abundance calls increases the ability to identify biomarkers and reduces the ambiguity of ion identifications. Continuing improvements in mass analyzer hardware have significantly increased mass resolution. Yet, these improvements come with an associated cost both for upgrading otherwise perfectly good capital equipment, and often in lost ion sensitivity. Software with the ability to quantitatively deconvolve nearly isobaric species from one another can extend the resolution and service life of any mass analyzer, without sacrificing sensitivity.
Citation:
Sokkalingam N, Ashrafi S, Schneider L, Tenderholt A, Wright W, Cajka T, Fiehn O, Lewitt M, Gerner L, Ramsey D, Schmidt D. PeakInvestigator™: software to improve the precision, resolution, and sensitivity of your mass analyzer. Poster presented at: 62nd Annual Am. Soc. Mass Spec.; 2014 Jun 15-19; Baltimore, MD.
Collaborations and Presentations
- UPenn, Epigenetics Institute, Affordable Proteomics collaboration presentation: June 2018
- UCSD, Herold lab, Hb Isotype Discovery by Isotopic Vector Analysis, collaboration presentation: MSACL, January 2018
- Metabolomics 2017 Showcase Challenge presentation: June 2017
- UPenn, Garcia Lab, Peptide Scoring Confidence collaboration: September 2016
- BYU, Price lab, inter-peak isotope spacing, collaboration presentation: September 2015
- UC Davis, Fiehn lab, Phase 2, collaboration presentation: May 2015
- UC Davis, Fiehn lab, Phase 1, collaboration presentation: September 2014
- Stanford, Snyder lab, collaboration presentation: May 2014