Sensory Lexicon Development and Validation for Gray Sufu Using Principal Component Analysis and Partial Mantel Tests

ABSTRACT Gray sufu, a subtype of Chinese fermented soybean curd, is a plant‐based protein product with cheese‐like characteristics. It is distinguished by its strong, distinctive aroma (often described as “stinky”) and gray surface coloration. Current research has primarily focused on microbial composition and flavor compounds in gray sufu, while comprehensive data on its physicochemical and sensory properties and their correlations remain limited. Therefore, this study aimed to develop and validate a sensory lexicon for profiling gray sufu. Twelve gray sufu samples and eight trained panelists were selected. Through free‐choice profiling, 93 descriptors were generated. After preliminary sorting, 58 descriptors were retained. Further reduction was achieved by calculating geometric means; descriptors with M % < 5% were eliminated, resulting in 30 descriptors. Three successive PCA iterations refined these to 22 and finally 13 descriptors. Validation via PCA and Partial Mantel tests of physicochemical and descriptor matrices confirmed that the 13‐descriptor lexicon could effectively characterize and differentiate gray sufu samples with minimal information loss. This investigation presents a novel approach by integrating physicochemical properties with multivariate statistical methods (PCA and Partial Mantel tests) for lexicon validation. Results provide practical guidance for sensory evaluation protocols and quality control in gray sufu production.