Rare breast cancer (BC) subtypes are associated with significantly poorer survival compared to other BC subtypes. The primary challenges in managing patients with rare BC subtypes include substantial variability in clinical behaviour and the absence of established molecular targets for treatment. This clinical challenge underscores the urgent need to investigate the mechanisms that drive tumour progression and metastasis, paving the way for novel therapeutic approaches. In the past decade, evidence has increasingly suggested that tumour heterogeneity is best characterised at the transcriptome level. To date, the transcriptomes of rare BC subtypes have been conducted at the bulk RNA level, obscuring critical cell-to-cell differences and cellular heterogeneity. Using the single-cell Fixed RNA Profiling assay on FFPE archival samples, we have profiled the transcriptomes of five rare BC subtypes known for their aggressive clinical behaviour. These include three estrogen-negative subtypes; metaplastic, medullary, and apocrine breast carcinomas and two estrogen-positive subtypes; micropapillary and pleomorphic lobular breast carcinomas. We have analysed 54 FFPE samples, yielding approximately 600,000 single cells. This extensive dataset has enabled us to construct a comprehensive single-cell atlas that maps the transcriptomic landscape of rare BC subtypes, providing valuable insights into its complexity and variability. Additionally, we are investigating the organisation of cells, ligand-receptor pairings, and cell-cell interactions within the tissue of rare BC subtypes through spatial transcriptomic analysis at single-cell resolution using the Xenium assay. Notably, due to the non-destructive nature of the Xenium workflow, we are able to perform H&E staining and immunofluorescence on the same section post-processing. This allows us to spatially register protein, histological, and RNA data into a unified image. Our scRNA data has identified diverse cell types and states, while the Xenium assay is being used to contextualise these findings within the tissue. Together, our study provides a robust framework for single-cell and spatial transcriptomics, enhancing our understanding of five rare distinct BC subtypes and aiding in identification of novel molecular drivers and druggable targets in these rare BC subtype.