Summary：**DeepLC 4.0.0b1 Beta Released: Exciting New Features Unveiled** *DeepLC: Retention time prediction

**DeepLC 4.0.0b1 Beta Released: Exciting New Features Unveiled**
*DeepLC: Retention time prediction for (modified) peptides using Deep Learning*

---

### Introduction
The proteomics community welcomed the beta launch of DeepLC 4.0.0b1 on November 2, 2025. This iteration builds on the established framework that predicts chromatographic retention times for both native and post‑translationally modified peptides using deep‑learning models. By refining algorithmic core components and expanding input flexibility, the developers aim to tighten the gap between theoretical predictions and experimental LC‑MS/MS observations, a critical step for high‑throughput biomarker discovery and targeted quantitation workflows.

### Key Developments
1. **Enhanced Model Architecture** – The beta introduces a hybrid CNN‑Transformer network that captures both local sequence motifs and long‑range dependencies, improving prediction accuracy for phosphorylated and glycosylated peptides by an average of 7 % over the previous version.
2. **Extended Modification Library** – Over 150 new chemical modifications, including labile adducts and isotopic labels, are now supported out‑of‑the‑box, reducing the need for custom parameter files.
3. **User‑Friendly Interface** – A streamlined command‑line wrapper and a lightweight GUI enable rapid batch processing, with built‑in visualization of prediction confidence intervals.
4. **Cross‑Platform Compatibility** – Pre‑compiled binaries for Windows, macOS, and Linux, plus Docker images, ensure reproducibility across diverse laboratory environments.
5. **Open‑Source Benchmark Suite** – The release ships with a curated dataset of 1.2 million peptide‑RT pairs from public repositories, facilitating community‑driven validation and model fine‑tuning.

### Industry Analysis
The beta arrives at a time when label‑free quantitation and data‑independent acquisition (DIA) dominate large‑scale proteomics projects. Accurate RT prediction directly influences peptide identification rates, especially for modified species that often elude conventional search engines. Early adopters report a reduction of false‑negative identifications by up