Biology subjects

Zeng, W.-F.

Publications and source records attributed to Zeng, W.-F..

15 recordsLinked to original sources

bioRxiv2026

FastaLake selects compact metaproteomic databases from large sequence reservoirs by de novo peptide lookup

Metaproteomics analyses require protein sequence databases, yet the microbial proteins present in complex communities are not known in advance. Searching reference catalogs containing hundreds of millions of candidate sequences reduces identification sensitivity, while generating sample-specific databases through metagenomic sequencing adds time and cost. We introduce FastaLake, a computational workflow that uses full-length de novo peptide predictions by AlphaNovo to retrieve complete proteins from broad sequence resources and selects compact databases for conventional searching. In a defined eight-member community, FastaLake approached reference-search peptide yields without knowing its members. Using public sequence catalogs, FastaLake identified about as many peptides as matched metagenomic databases in two stool cohorts and 20-35% more in intestinal fluid at the same nominal 1% peptide false discovery rate. The resulting proteomes reveal personal functional profiles that persist across intestinal regions despite differences in leading species between intestinal fluid and stool. DNA, RNA and protein profiles agreed on broad differences between patients but only weakly on individual functions. All 726 acquisition-specific databases were smaller than the human reference proteome. Compact databases allowed non-specific searches of all 347 CapScan acquisitions, revealing histidine enrichment at peptide C termini in intestinal fluid. FastaLake enables metaproteomics analysis without matched metagenomic sequencing by constructing search databases from sequence resources and de novo peptide sequencing.

bioinformatics↗

bioRxiv2026

AlphaPeptTools: scverse-native analysis of mass spectrometry-based proteomics

Mass spectrometry (MS)-based proteomics now routinely profiles proteomes at scale, but extracting biological insight and integrating complementary modalities remains challenging. Here, we present AlphaPeptTools, a Python package for the analysis of MS-proteomics data built on the AnnData data structure of the scverse. AlphaPeptTools implements scalable and modular strategies for MS-data ingestion, quality control, preprocessing, and statistical analysis while seamlessly integrating with the scverse ecosystem, unlocking multi-level, spatial, and multimodal analyses.

bioinformatics↗

bioRxiv2025

Tree-based quantification infers proteoform regulation in bottom-up proteomics data

Quantitative readout is essential in proteomics, yet current bioinformatics methods lack a framework to handle the inherent multi-level nature of the data (fragments, MS1 isotopes, charge states, modifications, peptides and genes). We present AlphaQuant, which introduces tree-based quantification. This approach organizes quantitative data into a hierarchical tree across levels. It allows differential analyses at fragment and MS1 level, recovering up to 50-fold more regulated proteins compared to a state-of-the-art approach. Using gradient boosting on tree features, we address the largely unsolved challenge of scoring quantification accuracy, as opposed to precision. Our method clusters peptides with similar quantitative behavior, providing a new approach to the protein grouping problem and enabling identification of regulated proteoforms directly from bottom-up data. Combined with deep learning classification, we infer phosphopeptides from proteome data alone, validating our findings with EGFR stimulation data. We then describe proteoform diversity across mouse tissues, revealing distinct patterns of post translational modifications and alternative splicing.

bioinformatics↗

bioRxiv2024

Carafe enables high quality in silico spectral library generation for data-independent acquisition proteomics

Data-independent acquisition (DIA)-based mass spectrometry is becoming an increasingly popular mass spectrometry acquisition strategy for carrying out quantitative proteomics experiments. Most of the popular DIA search engines make use of in silico generated spectral libraries. However, the generation of high-quality spectral libraries for DIA data analysis remains a challenge, particularly because most such libraries are generated directly from data-dependent acquisition (DDA) data or are from in silico prediction using models trained on DDA data. In this study, we developed Carafe, a tool that generates high-quality experiment-specific in silico spectral libraries by training deep learning models directly on DIA data. We demonstrate the performance of Carafe on a wide range of DIA datasets, where we observe improved fragment ion intensity prediction and peptide detection relative to existing pretrained DDA models. To make Carafe more accessible to the community, we have integrated Carafe into the widely used Skyline tool.

bioinformatics↗

bioRxiv2024

AlphaDIA enables End-to-End Transfer Learning for Feature-Free Proteomics

Mass spectrometry (MS)-based proteomics continues to evolve rapidly, opening more and more application areas. The scale of data generated on novel instrumentation and acquisition strategies pose a challenge to bioinformatic analysis. Search engines need to make optimal use of the data for biological discoveries while remaining statistically rigorous, transparent and performant. Here we present alphaDIA, a modular open-source search framework for data independent acquisition (DIA) proteomics. We developed a feature-free identification algorithm particularly suited for detecting patterns in data produced by sensitive time-of-flight instruments. It naturally adapts to novel, more eTicient scan modes that are not yet accessible to previous algorithms. Rigorous benchmarking demonstrates competitive identification and quantification performance. While supporting empirical spectral libraries, we propose a new search strategy named end-to-end transfer learning using fully predicted libraries. This entails continuously optimizing a deep neural network for predicting machine and experiment specific properties, enabling the generic DIA analysis of any post-translational modification (PTM). AlphaDIA provides a high performance and accessible framework running locally or in the cloud, opening DIA analysis to the community.

bioinformatics↗

bioRxiv2023

The potential of plasma HLA peptides beyond neoepitopes

Distinction of non-self from self is the major task of the immune system. Immunopeptidomics studies the peptide repertoire presented by the human leukocyte antigen (HLA) protein, usually on tissues. However, HLA peptides are also bound to plasma soluble HLA (sHLA), but little is known about their origin and potential for biomarker discovery in this readily available biofluid. Currently, immunopeptidomics is hampered by complex workflows and limited sensitivity, generally requiring several mL of plasma for the detection of hundreds of HLA peptides. Here, we take advantage of recent improvements in the throughput and sensitivity of mass spectrometry (MS)-based proteomics to develop a highly-sensitive, automated and economical workflow for HLA peptide analysis, termed Immunopeptidomics by Biotinylated Antibodies and Streptavidin (IMBAS). IMBAS-MS quantifies more than 5,000 HLA class I peptides from only 200 L of plasma, in just 30 minutes. Our technology revealed that the plasma immunopeptidome of healthy donors is remarkably stable throughout a year and strongly correlated between individuals with overlapping HLA types. Immunopeptides originating from diverse tissues, including the brain, are proportionately represented. We conclude that sHLAs are a promising avenue for immunology and precision oncology.

biochemistry↗

bioRxiv2023

Enhancing Inter-link Coverage in Cross-Linking Mass Spectrometry through Context-Sensitive Subgrouping and Decoy Fusion

In cross-linking mass spectrometry, sensitivity and specificity in assigning mass spectra to cross-links between different proteins (inter-links) remains challenging. Here, we report on limitations of commonly used concatenated target-decoy searches and propose a target-decoy competition strategy on a fused database as a solution. Further, we capitalize on context-divergent error rates by implementing a novel context-sensitive subgrouping strategy. This approach increases inter-link coverage by [~] 30 - 75 % across XL-MS datasets, maintains low error rates, and preserves structural accuracy.

biochemistry↗

bioRxiv2022

Robust dimethyl-based multiplex-DIA workflow doubles single-cell proteome depth via a reference channel

Single-cell proteomics aims to characterize biological function and heterogeneity at the level of proteins in an unbiased manner. It is currently limited in proteomic depth, throughput and robustness, a challenge that we address here by a streamlined multiplexed workflow using data-independent acquisition (mDIA). We demonstrate automated and complete dimethyl labeling of bulk or single-cell samples, without losing proteomic depth. In single runs of mammalian cells, a three-plex analysis of tryptic peptides quantified 7,700 proteins per channel. The Lys-N enzyme enables five-plex quantification at MS1 and MS2 level. Because the multiplex channels are quantitatively isolated from each other, mDIA accommodates a reference channel that does not interfere with the target channels. Our algorithm RefQuant takes advantage of this feature and confidently quantifies close to 4,000 proteins in single cells with excellent reproducibility, while our workflow currently allows routine analysis of 80 single cells per day. The concept of a stable proteome still holds at this deeper proteome coverage.

systems biology↗

bioRxiv2022

AlphaPeptDeep: A modular deep learning framework to predict peptide properties for proteomics

Machine learning and in particular deep learning (DL) are increasingly important in mass spectrometry (MS)-based proteomics. Recent DL models can predict the retention time, ion mobility and fragment intensities of a peptide just from the amino acid sequence with good accuracy. However, DL is a very rapidly developing field with new neural network architectures frequently appearing, which are challenging to incorporate for proteomics researchers. Here we introduce AlphaPeptDeep, a modular Python framework built on the PyTorch DL library that learns and predicts the properties of peptides (https://github.com/MannLabs/alphapeptdeep). It features a model shop that enables non-specialists to create models in just a few lines of code. AlphaPeptDeep represents post-translational modifications in a generic manner, even if only the chemical composition is known. Extensive use of transfer learning obviates the need for large data sets to refine models for particular experimental conditions. The AlphaPeptDeep models for predicting retention time, collisional cross sections and fragment intensities are at least on par with existing tools. Additional sequence-based properties can also be predicted by AlphaPeptDeep, as demonstrated with a novel HLA peptide prediction model to improve HLA peptide identification for data-independent acquisition.

bioinformatics↗

bioRxiv2022

AlphaViz: Visualization and validation of critical proteomics data directly at the raw data level

Although current mass spectrometry (MS)-based proteomics identifies and quantifies thousands of proteins and (modified) peptides, only a minority of them are subjected to in-depth downstream analysis. With the advent of automated processing workflows, biologically or clinically important results within a study are rarely validated by visualization of the underlying raw information. Current tools are often not integrated into the overall analysis nor readily extendable with new approaches. To remedy this, we developed AlphaViz, an open-source Python package to superimpose output from common analysis workflows on the raw data for easy visualization and validation of protein and peptide identifications. AlphaViz takes advantage of recent breakthroughs in the deep learning-assisted prediction of experimental peptide properties to allow manual assessment of the expected versus measured peptide result. We focused on the visualization of the 4-dimensional data cuboid provided by Bruker TimsTOF instruments, where the ion mobility dimension, besides intensity and retention time, can be predicted and used for verification. We illustrate how AlphaViz can quickly validate or invalidate peptide identifications regardless of the score given to them by automated workflows. Furthermore, we provide a predict mode that can locate peptides present in the raw data but not reported by the search engine. This is illustrated the recovery of missing values from experimental replicates. Applied to phosphoproteomics, we show how key signaling nodes can be validated to enhance confidence for downstream interpretation or follow-up experiments. AlphaViz follows standards for open-source software development and features an easy-to-install graphical user interface for end-users and a modular Python package for bioinformaticians. Validation of critical proteomics results should now become a standard feature in MS-based proteomics.

bioinformatics↗

bioRxiv2022

The structural context of PTMs at a proteome wide scale

The recent revolution in computational protein structure prediction provides folding models for entire proteomes, which can now be integrated with large-scale experimental data. Mass spectrometry (MS)-based proteomics has identified and quantified tens of thousands of post-translational modifications (PTMs), most of them of uncertain functional relevance. In this study, we determine the structural context of these PTMs and investigate how this information can be leveraged to pinpoint potential regulatory sites. Our analysis uncovers global patterns of PTM occurrence across folded and intrinsically disordered regions. We found that this information can help to distinguish regulatory PTMs from those marking improperly folded proteins. Interestingly, the human proteome contains thousands of proteins that have large folded domains linked by short, unstructured regions that are strongly enriched in regulatory phosphosites. These include well-known kinase activation loops that induce protein conformational changes upon phosphorylation. This regulatory mechanism appears to be widespread in kinases but also occurs in other protein families such as solute carriers. It is not limited to phosphorylation but includes ubiquitination and acetylation sites as well. Furthermore, we performed three-dimensional proximity analysis which revealed examples of spatial co-regulation of different PTM types and potential PTM crosstalk. To enable the community to build upon these first analyses, we provide tools for 3D visualization of proteomics data and PTMs as well as python libraries for data accession and processing.

systems biology↗

bioRxiv2021

AlphaPept, a modern and open framework for MS-based proteomics

In common with other omics technologies, mass spectrometry (MS)-based proteomics produces ever-increasing amounts of raw data, making their efficient analysis a principal challenge. There is a plethora of different computational tools that process the raw MS data and derive peptide and protein identification and quantification. During the last decade, there has been dramatic progress in computer science and software engineering, including collaboration tools that have transformed research and industry. To leverage these advances, we developed AlphaPept, a Python-based open-source framework for efficient processing of large high-resolution MS data sets. Using Numba for just-in-time machine code compilation on CPU and GPU, we achieve hundred-fold speed improvements while maintaining clear syntax and rapid development speed. AlphaPept uses the Python scientific stack of highly optimized packages, reducing the code base to domain-specific tasks while providing access to the latest advances in machine learning. We provide an easy on-ramp for community validation and contributions through the concept of literate programming, implemented in Jupyter Notebooks of the different modules. A framework for continuous integration, testing, and benchmarking enforces solid software engineering principles. Large datasets can rapidly be processed as shown by the analysis of hundreds of cellular proteomes in minutes per file, many-fold faster than the data acquisiton. The AlphaPept framework can be used to build automated processing pipelines using efficient HDF5 based file formats, web-serving functionality and compatibility with downstream analysis tools. Easy access for end-users is provided by one-click installation of the graphical user interface, for advanced users via a modular Python library, and for developers via a fully open GitHub repository.

bioinformatics↗

bioRxiv2021

Precise, Fast and Comprehensive Analysis of Intact Glycopeptides and Monosaccharide-Modifications with pGlyco3

We present a glycan-first glycopeptide search engine, pGlyco3, to comprehensively analyze intact N- and O-glycopeptides, including glycopeptides with modified saccharide units. A novel glycan ion-indexing algorithm developed in this work for glycan-first search makes pGlyco3 5-40 times faster than other glycoproteomic search engines without decreasing the accuracies and sensitivities. By combining electron-based dissociation spectra, pGlyco3 integrates a fast, dynamic programming-based algorithm termed pGlycoSite for site-specific glycan localization (SSGL). Our evaluation based on synthetic and natural glycopeptides showed that the SSGL probabilities estimated by pGlycoSite were proved to be appropriate to localize site-specific glycans. With pGlyco3, we found that N-glycopeptides and O-mannose glycopeptides in yeast samples were extensively modified by ammonia adducts on Hex (aH) and verified the aH-glycopeptide identifications based on released N-glycans and 15N/13C-labeled data. Thus pGlyco3, which is freely available on https://github.com/pFindStudio/pGlyco3/releases, is an accurate and flexible tool to identify glycopeptides and modified saccharide units.

bioinformatics↗

bioRxiv2020

pDeep3: Towards More Accurate Spectrum Prediction with Fast Few-Shot Learning

Spectrum prediction using deep learning has attracted a lot of attention in recent years. Although existing deep learning methods have dramatically increased the pre-diction accuracy, there is still considerable space for improvement, which is presently limited by the difference of fragmentation types or instrument settings. In this work, we use the few-shot learning method to fit the data online to make up for the shortcoming. The method is evaluated using ten datasets, where the instruments includes Velos, QE, Lumos, and Sciex, with collision energies being differently set. Experimental results show that few-shot learning can achieve higher prediction accuracy with almost negligible computing resources. For example, on the dataset from a untrained instrument Sciex-6600, within about 10 seconds, the prediction accuracy is increased from 69.7% to 86.4%; on the CID (collision-induced dissociation) dataset, the prediction accuracy of the model trained by HCD (higher energy collision dissociation) spectra is increased from 48.0% to 83.9%. It is also shown that, the method is not critical to data quality and is sufficiently efficient to fill the accuracy gap. The source code of pDeep3 is available at http://pfind.ict.ac.cn/software/pdeep3.

bioinformatics↗

bioRxiv2018

Open-pFind enables precise, comprehensive and rapid peptide identification in shotgun proteomics

Shotgun proteomics has grown rapidly in recent decades, but a large fraction of tandem mass spectrometry (MS/MS) data in shotgun proteomics are not successfully identified. We have developed a novel database search algorithm, Open-pFind, to efficiently identify peptides even in an ultra-large search space which takes into account unexpected modifications, amino acid mutations, semi- or non-specific digestion and co-eluting peptides. Tested on two metabolically labeled MS/MS datasets, Open-pFind reported 50.5-117.0% more peptide-spectrum matches (PSMs) than the seven other advanced algorithms. More importantly, the Open-pFind results were more credible judged by the verification experiments using stable isotopic labeling. Tested on four additional large-scale datasets, 70-85% of the spectra were confidently identified, and high-quality spectra were nearly completely interpreted by Open-pFind. Further, Open-pFind was over 40 times faster than the other three open search algorithms and 2-3 times faster than three restricted search algorithms. Re-analysis of an entire human proteome dataset consisting of [~]25 million spectra using Open-pFind identified a total of 14,064 proteins encoded by 12,723 genes by requiring at least two uniquely identified peptides. In this search results, Open-pFind also excelled in an independent test for false positives based on the presence or absence of olfactory receptors. Thus, a practical use of the open search strategy has been realized by Open-pFind for the truly global-scale proteomics experiments of today and in the future.

bioinformatics↗