Extended Data Table 1. A lesion segregation based test for oncogenic selection.

Strain	Gene	Mutation	Mutation count	Odds ratio	P-value	Known driver
C3H	Braf	6:37548568_A/T	151	2.13	5.77x10-6	Yes
C3H	Hras	7:145859242_T/C	81	2.67	6.88x10-6	Yes
C3H	Hras	7:145859242_T/A	65	1.02	1	Yes
C3H	Intronic Fmnl1	11:105081902_A/C	44	1.03	1	No
C3H	Intergenic	9:73125689_G/C	42	1.13	1	No
C3H	Egfr	11:14185624_T/A	34	3.87	1.23x10-4	Yes
CAST	Braf	6:37451282_A/T	42	1.41	0.338	Yes

Recurrently mutated sites in both C3H and CAST with sufficient estimated power to detect oncogenic selection through biased strand retention analysis (required >33 C3H recurrences or >41 CAST recurrences). Odds ratio values >1 indicate the predicted correlation of driver mutation and Watson/Crick strand retention in tumours with the candidate driver mutation, but not for those without the mutation. The Fisher’s exact test P-value is shown after Bonferroni correction. Known driver indicates the mutation or its orthologous change has previously been implicated as a driver of hepatocellular carcinoma⁶. The CAST 6:37451282_A/T mutation is orthologous to the C3H 6:37548568_A/T mutation.