Author Archives: Vera

Development of Machine Learning assisted Fingermark Imaging Software (iFIS)

June 2026, ASMS Conference

Simona Francese¹; Elias Jensen²; Sara Tortorella³; Chloe Spencer^{1, 4}; Giuseppe Arturi⁵; Simon Cross⁶; Hassan Ugail²

¹Sheffield Hallam University, Sheffield, United Kingdom; ²University of Bradford, Bradford, United Kingdom; ³Mass Analytica, Sant Cugat del Valles, Spain; ⁴University of Nottingham, Nottingham, United Kingdom; ⁵Molecular Discovery, Borehamwood, United Kingdom; ⁶Mass Analytica, S.L., Borehamwood, United Kingdom

 

Abstract

Introduction

Molecular fingerprinting has been featured in the Fingermark Visualisation Manual edited by Dstl/Home Office and is being used in Police casework. It encompasses the application of Mass Spectrometry Imaging (MSI), particularly MALDI MSI, for the provision of biometric information, through generating multiple molecular images of crime scene fingermark evidence, alongside contextual (molecular) information. Whilst both advanced freeware and proprietary MSI exist, they are complex as mostly built to process biological tissue imaging data. We have developed a dedicated software, enabling auditable fingermark images “manipulation”, seamless separation of overlapping fingermarks and, crucially, integrating a machine learning algorithm capable of grading and finding fingermark images of the highest quality within a seconds, without manually having to inspect the entire mass range.

Methods

Following Ethical approval (ER52762288), fingermarks were matrix spray-coated using the HTX M3+™ Sprayer (HTX Technologies, USA) and imaged on a SELECT SERIES MRT MALDI mass spectrometer (Waters Corporation, UK) in positive mode. Lipostar MSI (Molecular Discovery, UK) – used as the skeleton to build iFIS –  generated 663 images which were graded according to the 5-point scale Scotland Yard system machine learning model. The ML algorithm utilised a range of deep features based on the Resnet50 architecture corresponding to the visual characteristics of the fingermark (minutiae and texture features). These were then fed to a Support Vector Machine-based classification algorithm for categorizing fingermarks into five distinct categories.

You must be logged in to access this content. Not yet registered? Create a new account

From Molecular Structure to Pharmacokinetic Parameters: Autonomizing Quantitative Bioanalysis Across Modalities

June 2026, ASMS Conference

Ismael Zamora^{1, 2}; Luca Morettoni³; Fabien Fontaine²; Kevin Bateman⁴

¹Mass Spec Analytica, Sant Cugat del Valles, Spain; ²Lead Molecular Design, SL, Sant Cugat del Vallès, Spain; ³Mass Analytica, Sant Cugat del Valles, Spain; ⁴2KDAM Consulting, Halifax, NS

 

Abstract

Introduction

Mass spectrometry based quantitative analysis, typically built on triple quadrupole (QQQ) instruments, requires a disproportionate number of manual procedures for such an advanced analytical technique. The use of high-resolution mass spectrometry (HRMS) with data independent acquisition can obviate the need for compound specific methods but this approach lacks the required data processing capabilities to fully leverage the data content. We describe the development of a platform that automates all aspects of data processing through data reporting of pharmacokinetic parameters. The user provides the structure and assay acceptance criteria, and the platform holistically finds a quantitative model that best represents the data. We demonstrate this platform for drug discovery pharmacokinetic studies across multiple modalities, including small molecules, peptides and antisense oligonucleotides.

Methods

Plasma samples from rat PK studies of Sitagliptin, Liraglutide and  Mipomersen were processed for LC-MS analysis.  A Sciex ZenoTOF 7600 quadrupole time of flight mass spectrometer was used for data collection. A SWATH acquisition method with a TOF MS scan followed by MSMS scans collected using 25 Da windows was used for data acquisition. Data processing was done using AI-Quant (described below).

You must be logged in to access this content. Not yet registered? Create a new account

End-to-End High-Throughput Biotransformation Workflow: Automated Data Acquisition and Processing of Sub-Second UHPLC Peaks Using Multi-Reflecting Time-of-Flight Mass Spectrometry and Data-Mining

June 2026, ASMS Conference

Ismael Zamora¹; Hania Khoury-Hollins²; Richard Lock²; David Pickles²; Robert S Plumb²; Ian Wilson³

¹Mass Spec Analytica, Sant Cugat del Valles, Spain; ²Waters Corporation, Wilmslow, United Kingdom; ³Imperial College London, London, United Kingdom

 

Abstract

Introduction

Traditional ultra-high performance liquid chromatography (UHPLC™) paired with high-resolution mass spectrometry (HRMS) has long been the standard for drug metabolite characterization, but its throughput is limited by peak dispersion the need to balance mass resolution with scan speed, and data processing and rationalisation. These factors, along with labour-intensive manual data review, slow down the delivery of actionable results.

To overcome these bottlenecks, we introduce a fully integrated high-throughput data dependent workflow combining a next generation UHPLC with multi reflecting time-of-flight technology (Xevo™ MRT MS) and dedicated data-mining software. This platform enables rapid, automated acquisition and processing of sub-second UHPLC peaks, achieving part-per-billion mass accuracy (≤500 ppb RMS), scan speeds up to 100 Hz, and high mass resolution (100K FWHM).

Methods

Male beagle dogs received a single intravenous (IV) dose, and whole blood was collected from the jugular vein both before and after dosing. Plasma was obtained by centrifugation and analyzed using reversed-phase UHPLC coupled to multi reflecting time of flight mass spectrometer.

Data dependent acquisition mode was used, and metabolites were detected and quantified in both positive and negative electrospray ionization modes, utilizing Xevo MRT MS and a dedicated MassMetaSite software.

Workflow improvements were assessed by comparing results obtained with shorter runtimes to those from conventional UHPLC method used by the contract research organization (CRO) that prepared the samples. This comparison focused on data quality, sample analysis times, and confidence in structural characterization.

You must be logged in to access this content. Not yet registered? Create a new account

Uncertainty-Aware Site-of-Metabolism Prediction from Ambiguous LC–MS Metabolite Identification Data

June 2026, ASMS Conference

Ramon Adàlia^{1, 2}; Ismael Zamora³

¹Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Spain; ²Lead Molecular Design, SL, Sant Cugat del Vallès, Spain; ³Mass Analytica, Sant Cugat del Valles, Spain

 

Abstract

Introduction

Liquid chromatography–mass spectrometry (LC–MS) is the dominant technology for metabolite identification in early drug discovery, yet it frequently produces structurally ambiguous metabolites with multiple plausible sites of metabolism (SoMs). Although this ambiguity is well understood in LC–MS workflows, most computational SoM models require unambiguous, binary annotations and therefore cannot directly exploit discovery-stage data. As a result, a large fraction of routinely generated LC–MS metabolite information is excluded from predictive modeling. We present an uncertainty-aware modeling strategy that preserves LC–MS-derived structural ambiguity by encoding relative SoM plausibility, enabling direct use of metabolite identification data without requiring definitive structure elucidation.

Methods

Human liver microsome LC–MS metabolite identification data were processed with software to generate candidate metabolite structures consistent with observed mass shifts and fragmentation patterns. For each metabolite peak, atom-level soft labels were constructed by averaging SoM assignments across equally scoring structural hypotheses, yielding relative plausibility scores rather than binary labels. Metabolite peaks assigning nonzero labels to an excessively large fraction of atoms were filtered during label construction to control noise. Atom rankings were learned using a graph attention neural network trained with a pairwise ranking objective. Molecular graph features were augmented with atom-level reactivity scores from MetaSite7. Model performance was evaluated using ranking-based metrics on the soft-labeled dataset and top-2 accuracy on an independent benchmark with experimentally confirmed SoMs.

You must be logged in to access this content. Not yet registered? Create a new account