Fig. 3: Identification and transcriptional characterization of premalignant thyrocytes.

a t-SNE projection of the two clusters of non-malignant thyrocytes (c01 and c02). b Potential trajectory of c01 and c02 cells inferred by Monocle2. The arrow shows the putative developmental direction from c01 to c02 cluster. c Violin plots showing several DEGs between c01, c02, and c03-c09 clusters. Expression level of TG, TPO, and IYD decreases while the level of TMSB4X increases continuously from c01 to c02 and to c03-c09 thyrocytes. d Representative H&E staining of primary tumor (T5) and para-tumor (P5) of Case 5 at magnifications of × 20, × 100, and × 400. Three independent experiments were performed and generated similar results. Scale bar (upper) = 300 µm; Scale bar (middle) = 100 µm; Scale bar (bottom) = 50 µm. e Boxplot showing TDS scores of thyrocytes from the premalignant paratumor (n = 1036), malignant paratumor (n = 118), tumor (n = 1337) and lymph node metastasis (n = 4986) of Case 5. The middle lines of the boxplots show the median (central line), the lower and upper hinges show the 25–75% interquartile range (IQR), and the whiskers extend from the hinge to the farthest data point within a maximum of 1.5 x IQR. f Trajectory analysis of four thyrocyte populations from Case 5 (P5 premalignant cells, P5/T5/LN5r malignant cells). The arrow shows potential evolutionary direction from P5 premalignant thyrocytes to P5/T5/LN5r malignant thyrocytes. g Heatmap of normalized expression of top upregulated and downregulated genes between normal (c01) and premalignant thyrocytes (c02). Source data are provided in the Source data file.