Serval: A modular framework for decoding imaging based spatial transcriptomics data.

Serval's modular framework and CoSiD decoder significantly enhance transcript recovery and biological structure annotation across multiple spatial transcriptomics platforms, including MERFISH and DART-FISH.

• Why it matters: Accurate decoding of fluorescence patterns into gene identities is crucial for reliable spatial transcriptomics analysis, yet systematic benchmarking of decoding methods has been limited, hindering progress.
• What they did: The authors developed Serval, a flexible, modular framework for decoding, and introduced CoSiD, a cosine similarity-based decoder, evaluating their performance on synthetic and real datasets from different platforms.
• The result: CoSiD outperformed existing methods by increasing transcript recovery and clustering stability, enabling more precise biological insights, and demonstrating the framework's platform-agnostic versatility.