MSACL 2018 US Abstract

Topic: Data Science

Statistical Identification and Quantitative Deconvolution of Hemoglobin Beta Chain Isotypes Unresolved by LC-MS by Isotopic Vector Angle Analysis

Luke Schneider (Presenter)
Target Discovery, Inc./ Veritomyx, Inc.

Bio: Dr. Schneider is the founder and Chief Scientific Officer of Target Discovery, Inc. Target Discovery is a diagnostics company developing assays based on protein isoforms for oncology. Dr. Schneider is co-inventor of the Mass Defect Tagging proteomics method, the EOtrol line of dynamic coatings for capillary electrophoresis, and the TD buffer system (sold through Pressure Biosciences, Inc.) that support intact membrane protein recovery from tissue samples in a format suitable to immunoaffinity enrichment/mass spectrometry and ELISA. Veritomyx, Inc. (a subsidiary of Target Discovery) is commercializing software to support statistical analysis of mass spectrometric data. Dr. Schneider has been in diagnostics for more than 30 years previously serving as the Upconverting Phosphor Diagnostics Center Director at SRI International.

Authorship: Luke V. Schneider(1), Nandhini Sokkalingam(1), David A. Herold(2), Jane Y. Yang(2)
(1) Target Discovery, Inc., Palo Alto, CA, (2) University of California, San Diego, CA

Short Abstract

Isotopic Vector Angle (IVA) analysis is a new approach to statistically detect and quantitatively deconvolve nearly complete overlapping isotopic patterns within one spectrum. This method is based on the computational representation of isotopic abundance patterns as vectors and comparison between patterns via calculation of vector angles. In this paper, IVA analysis statistically detects and deconvolves chromatographically unresolved hemoglobin variants with a mass difference of less than 4 mass units from normal hemoglobin beta chain.

Long Abstract


There are over 1200 reported hemoglobin (Hgb) variants, a majority of which are SNPs (single nucleotide polymorphisms) that manifest in a single amino acid substitution. Current techniques used to detect these variants, such as ion exchange HPLC and capillary electrophoresis (CE) are limited to identifying the variant hemoglobin based on reference data for retention times, migration zones, and relative concentration. In these cases, HPLC-MS to determine the mass of the Hgb variant can assist in identification. However, if normal and variant Hgb chains are chromatographically unresolved and the mass shift is < 6 mass units, even HPLC-MS is of limited utility to determine the identity of the variant. Herein, we present a new method Isotopic Vector Angle (IVA) analysis to statistically identify and quantify Hgb variants that have a nearly complete mass overlap with normal Hgb from one mass spectrum.


The LC/MS dataset used for this study includes seven patients, six with hemoglobinopathies confirmed by cation exchange HPLC, CE, and LC-MS. Three spectra came from one patient with normal Hgb beta chain (15,867 Da). The other six patients exhibited Hgb variants in the beta chain along with normal. In one of the patients (T), the variant co-eluted with the normal Hgb and was unresolved based on mass. Isotopic relative abundance vectors were constructed from the 18+ and 19+ charge states of hormal and variant Hgb beta chains by averaging the abundances of 11 consecutive scans. The resulting isotopic abundance vectors were compared to each other by calculating the vector angle between them. The 95% confidence limits of each isotopic abundance vectors were also determined and translated into isotopic vector angle confidence limits so that standard statistical methods could be applied to the resulting isotopic vector angles.


All the patient samples exhibited the same Isotopic Vector Angles within 95% confidence, except patient T. The IVA for the Hgb beta chain (at 6.3 min) for patient T was found to be statistically higher than zero (a perfect match) with a p-score < 0.001 (Student’s t-test, 1-tailed). For validation of the method we used the normal (wild type) Hgb alpha chain peaks at the 8 min elution time, none of these Hgb alpha chain samples were statistically differentiated from the standard vector at this position. Assuming that the Hgb beta chain variant differed by only a single amino acid substitution, we represented its isotopic abundance vector by the that of the normal Hgb beta chain, only zero charge mass shifted by -4 to +2 Da. It was then possible to solve for the optimal linear combination of isotopic vectors that best matched the measured isotopic Hgb beta pattern of patient (T) by minimization of the isotopic vector angle for different mass shifts of the two vectors. The best match occurring at a -1 Da mass shift (corresponding to a 59% variant composition and the variant most likely being K>E, I>N, Q>E, or N>E based on mass differences of amino acid substitutions allowed by SNPs.

Conclusions & Discussion

This study shows that IVA analysis can be used to detect even subtle changes in a mass spectrometric isotopic pattern from an expected pattern and can do this with statistical confidence. Where the isotopic patterns of two (or more) species overlap, this study shows that it is possible to quantitatively deconvolve the relative contributions of these overlapped species using isotopic vector angle analysis. The uncertainty in the measured isotopic pattern can also be used to estimate the confidence intervals for both detection and deconvolved abundances.

References & Acknowledgements:

1. Schneider, L.V., “Isotopic Vector Angle Analysis”: see this white paper, DOI: 10.13140/RG.2.2.31392.69120.

2. Schneider, L.V., N. Sokkalingam, D.A. Herold, J.Y. Yang, "Isotopic Vector Angle Analysis to Identify Protein Isoforms in Mass Spectrometry", poster presented at Intl HUPO conference, Sep 17-20, Dublin, Ireland.

Financial Disclosure

SalaryyesTarget Discovery, Inc.
Board MemberyesTarget Discovery, Inc. and Aromyx, Inc.
Stockyes Target Discovery, Inc. and Aromyx, Inc.

IP Royalty: no

Planning to mention or discuss specific products or technology of the company(ies) listed above: