Abstract
MOTIVATION: Spatial transcriptomics (ST) technologies provide valuable insights into cellular heterogeneity by simultaneously acquiring both gene expression profiles and cellular location information. However, the limited diversity and accuracy of "gold standard" datasets hindered the effectiveness and fairness of benchmarking rapidly growing ST analysis tools. RESULTS: To address this issue, we proposed Spider, a flexible and comprehensive framework for simulating ST data without requiring real ST data as a reference. By characterizing the spatial patterns using cell type proportions and transition matrix between adjacent cells, Spider can produce more realistic and diverse simulated data and offer enhanced modeling flexibility compared to existing simulation methods. Additionally, Spider provides interactive features for customizing the spatial domain, such as zone segmentation and integration of histology imaging data. Benchmark analyses demonstrate that Spider outperforms other simulation tools in preserving the spatial characteristics of real ST data and facilitating the evaluation of downstream analysis methods. Spider is implemented in Python and available at https://github.com/YANG-ERA/Spider. AVAILABILITY AND IMPLEMENTATION: All codes, simulated ST data in this paper are publicly available at https://github.com/YANG-ERA/Spider.