54Understanding circuit organization depends on identification of cell types. Recent advances in 55 transcriptional profiling methods have enabled classification of cell types by their gene 56 expression. While exceptionally powerful and high throughput, the ground-truth validation of 57 these methods is difficult: if cell type is unknown, how does one assess whether a given 58 analysis accurately captures neuronal identity? To shed light on the capabilities and limitations 59 of solely using transcriptional profiling for cell type classification, we performed two forms of 60 transcriptional profiling -RNA-seq and quantitative RT-PCR, in single, unambiguously identified 61 neurons from two small crustacean networks: the stomatogastric and cardiac ganglia. We then 62 combined our knowledge of cell type with unbiased clustering analyses and supervised machine 63 learning to determine how accurately functionally-defined neuron types can be classified by 64 expression profile alone. Our results demonstrate that expression profile is able to capture 65 neuronal identity most accurately when combined with multimodal information that allows for 66 post-hoc grouping so analysis can proceed from a supervised perspective. Solely unsupervised 67 clustering can lead to misidentification and an inability to distinguish between two or more cell 68 types. Therefore, our study supports the general utility of cell identification by transcriptional 69 profiling, but adds a caution: it is difficult or impossible to know under what conditions 70 transcriptional profiling alone is capable of assigning cell identity. Only by combining multiple 71 modalities of information such as physiology, morphology or innervation target can neuronal 72 identity be unambiguously determined. 73 74 3
SIGNIFICANCE STATEMENT 75Single cell transcriptional profiling has become a widespread tool in cell identification, 76 particularly in the nervous system, based on the notion that genomic information determines cell 77 identity. However, many cell type classification studies are unconstrained by other cellular 78 attributes (e.g., morphology, physiology). Here, we systematically test how accurately 79 transcriptional profiling can assign cell identity to well-studied anatomically-and functionally-80 identified neurons in two small neuronal networks. While these neurons clearly possess distinct 81 patterns of gene expression across cell types, their expression profiles are not sufficient to 82 unambiguously confirm their identity. We suggest that true cell identity can only be determined 83 by combining gene expression data with other cellular attributes such as innervation pattern, 84 morphology, or physiology. 85 86 87 88 89Unambiguous classification of neuronal cell types is a long-standing goal in 90 neuroscience with the aim to understand the functional components of the nervous system that 91give rise to circuits and, ultimately, behavior (1-6). Beyond that, agreement upon neuronal cell 92 types provides the opportunity to greatly increase reproduci...