This preprint introduces a computational method called 'weighted-nearest neighbor' (WNN) analysis for integrating multimodal single-cell data. The method enables the identification of cell states based on multiple data types, including RNA and protein expression, chromatin accessibility, DNA methylation, and spatial location. The authors applied WNN to a CITE-seq dataset of 211,000 human peripheral blood mononuclear cells (PBMCs) with 228 antibodies, generating a multimodal reference atlas of the circulating immune system. They demonstrated that WNN analysis significantly improves the resolution of cell states and the identification of previously unreported lymphoid subpopulations. The method was also used to rapidly map new datasets and interpret immune responses to vaccination and SARS-CoV-2 infection. The authors implemented WNN in an updated version of their open-source R toolkit Seurat, making it broadly applicable for integrative multimodal analysis of single-cell data. The study highlights the importance of multimodal integration in overcoming the limitations of single-modality approaches and provides a flexible framework for analyzing diverse datasets. The results show that WNN analysis outperforms alternative methods in terms of accuracy and speed, and that it can be applied to various multimodal technologies, including ATAC-seq and ASAP-seq. The authors also demonstrated the utility of their approach in characterizing immune cell states, identifying differentially expressed markers, and mapping query datasets to the multimodal reference. The study concludes that WNN analysis provides a robust and flexible strategy for integrative multimodal analysis of single-cell data, enabling a unified definition of cellular identity and function. The data and code are available for public use.This preprint introduces a computational method called 'weighted-nearest neighbor' (WNN) analysis for integrating multimodal single-cell data. The method enables the identification of cell states based on multiple data types, including RNA and protein expression, chromatin accessibility, DNA methylation, and spatial location. The authors applied WNN to a CITE-seq dataset of 211,000 human peripheral blood mononuclear cells (PBMCs) with 228 antibodies, generating a multimodal reference atlas of the circulating immune system. They demonstrated that WNN analysis significantly improves the resolution of cell states and the identification of previously unreported lymphoid subpopulations. The method was also used to rapidly map new datasets and interpret immune responses to vaccination and SARS-CoV-2 infection. The authors implemented WNN in an updated version of their open-source R toolkit Seurat, making it broadly applicable for integrative multimodal analysis of single-cell data. The study highlights the importance of multimodal integration in overcoming the limitations of single-modality approaches and provides a flexible framework for analyzing diverse datasets. The results show that WNN analysis outperforms alternative methods in terms of accuracy and speed, and that it can be applied to various multimodal technologies, including ATAC-seq and ASAP-seq. The authors also demonstrated the utility of their approach in characterizing immune cell states, identifying differentially expressed markers, and mapping query datasets to the multimodal reference. The study concludes that WNN analysis provides a robust and flexible strategy for integrative multimodal analysis of single-cell data, enabling a unified definition of cellular identity and function. The data and code are available for public use.