TABLE 1.

SVM model predicts tumour classification.

Sample	Histology	BRAF/FGFR1	Consensus clustering	SVM
VAL1	GG	BRAF V600E	Group 1	Group 1
VAL2	GG	BRAF V600E	Group 1	Group 1
VAL3	GG	‐	Group 1	Group 1
VAL4	GG	‐	Group 1	Group 1
VAL5	GG	‐	Group 1	Group 1
VAL6	GG	‐	Group 1	Group 1
VAL7	GNT NOS	BRAF V600E	Group 1	Group 1
VAL8	GNT NOS	‐	Group 1	Group 1
VAL34	DNET	‐	Group 2	Group 1*
VAL16	DNET	FGFR1‐TACC1	Group 2	Group 2
VAL17	DNET	FGFR1 TKD	Group 2	Group 2
VAL18	DNET	FGFR1 TKD	Group 2	Group 2
VAL19	DNET	FGFR1 TKD	Group 2	Group 2
VAL20	DNET	FGFR1 TKD	Group 2	Group 2
VAL21	DNET	FGFR1 TKD	Group 2	Group 2
VAL22	DNET	FGFR1 TKD	Group 2	Group 2
VAL23	DNET	FGFR1 TKD	Group 2	Group 2
VAL24	DNET	FGFR1 TKD	Group 2	Group 2
VAL25	DNET	FGFR1 TKD	Group 2	Group 2
VAL26	DNET	‐	Group 2	Group 2
VAL27	DNET	‐	Group 2	Group 2
VAL28	GNT NOS	‐	Group 2	Group 2
VAL29	GNT NOS	‐	Group 2	Group 2
VAL10	GG	BRAF V600E	Control/Diffuse	Group 1
VAL11	GG	BRAF V600E	Control/Diffuse	Group 1
VAL12	GG	BRAF V600E	Control/Diffuse	Group 1
VAL13	GG	BRAF V600E	Control/Diffuse	Group 1
VAL32	GNT NOS	‐	Control/Diffuse	Group 1
VAL33	GG	‐	Control/Diffuse	Group 1
VAL9	GG	BRAF V600E	Control/Diffuse	Group 1
VAL15	GG	BRAF V600E	Control/Diffuse	Group 2
VAL14	GG	BRAF V600E	Control/Diffuse	Control
VAL30	DNET	FGFR1 TKD	Control/Diffuse	Control
VAL31	GG	‐	Control/Diffuse	Control

Note: Twenty‐three tumours belonging to the validation cohort clustered alongside the training cohort by consensus clustering. The support vector model (SVM) predicted concordant classification in 22/23 (96%).

^*A single misclassification in a sample with >10% array probe failure.