{"version":1,"type":"story","url":"https://digestai.news/story/researchers-present-proformer-for-classifying-proteomic-data","json":"https://digestai.news/story/researchers-present-proformer-for-classifying-proteomic-data.json","markdown":"https://digestai.news/story/researchers-present-proformer-for-classifying-proteomic-data.md","slug":"researchers-present-proformer-for-classifying-proteomic-data","headline":"Researchers present ProFormer for classifying proteomic data","summary":"A research team from the University of Heidelberg has introduced ProFormer, a deep learning pipeline designed to classify biological samples directly from mass spectrometry data. The model uses a transformer architecture to process tabular MS1-level features derived from peptide ions, bypassing the need for time-consuming manual data interpretation. The authors state that ProFormer outperforms traditional machine learning methods and image-based convolutional neural networks across 14 evaluated architectures.\n\nThe study demonstrates the model's versatility by applying it to diverse liquid chromatography-mass spectrometry datasets. ProFormer successfully classified single-cell proteomes by cell type, cycle stage, or differentiation trajectory. It also accurately determined patient disease status from plasma proteomes. The researchers highlight that the model preserves signal contribution at single-peptide resolution, enabling explainable AI insights into how the system aggregates data.\n\nThis work addresses the limited throughput of current proteomics technologies, which restricts the robust classification of large clinical cohorts. By providing a rapid framework for proteomic data classification, the tool has implications for patient stratification, early disease detection, and single-cell analysis. The project was funded by the Heidelberg Explorer Call, the Federal Ministry of Education and Research, and the Ministry of Science Baden-Württemberg.","keyPoints":["ProFormer is a transformer pipeline that classifies samples using tabular MS1-level features from peptide ions.","The model outperforms traditional machine learning and CNNs across 14 evaluated architectures in classification tasks.","ProFormer accurately classified single-cell proteomes by cell type and patient disease status from plasma proteomes."],"whyItMatters":"ProFormer offers a faster, more generalizable method for analyzing proteomic data, potentially accelerating biomarker discovery and patient stratification in clinical settings by removing manual interpretation steps.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":["University of Heidelberg"],"models":["ProFormer"],"people":["Krull, K.K.","Kühn, A.","Höhn, J."]},"firstPublishedAt":"2026-10-05T00:00:00Z","updatedAt":"2026-10-05T00:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"Nature Machine Learning","title":"ProFormer: generalizable classification of single-cell and plasma proteomes using deep learning","url":"https://nature.com/articles/s41467-026-78143-2","publishedAt":"2026-10-05T00:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"Researchers present ProFormer for classifying proteomic data\", 5 October 2026, https://digestai.news/story/researchers-present-proformer-for-classifying-proteomic-data","publisher":"Digest AI","title":"Researchers present ProFormer for classifying proteomic data","datePublished":"2026-10-05T00:00:00Z","url":"https://digestai.news/story/researchers-present-proformer-for-classifying-proteomic-data"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}