Researchers have developed PLATO, a high-throughput platform designed to map the spatial proteome across entire tissue sections with high resolution. Traditional methods often struggle with high costs, low efficiency, or a limited number of detectable proteins, but this new approach uses microfluidic chips to perform parallel-flow sampling from multiple angles. To transform these captured measurements into detailed images, the team created Flow2Spatial, a deep learning algorithm that utilizes transfer learning from other omics data to reconstruct protein distributions. The study demonstrates the effectiveness of this system by successfully mapping complex structures in the mouse cerebellum, rat intestinal villi, and human breast cancer tissues. By identifying distinct tumor subtypes and molecular changes during inflammation, PLATO provides a powerful tool for understanding tissue heterogeneity in both fundamental biology and clinical diagnostics. This framework significantly reduces the number of required mass spectrometry measurements while maintaining subcellular-scale insights across large tissue areas.
References:
- Hu B, He R, Pang K, et al. High-resolution spatially resolved proteomics of complex tissues based on microfluidics and transfer learning[J]. Cell, 2025, 188(3): 734-748. e22.
