Supplementary Figure 1 A B C

Supplementary Figure 1 Supplementary Figure 1. Mutational signature of AF-HCC. (A) Percentage of the six possible mutation classes in the whole genome of AF-HCC. (B) Strand bias in the whole exome of the six mutation classes in AF-HCC and TCGA-HCC. (C) The context of C>A mutations in coding regions of AF-HCC. For example, C>A:GCC indicates C>A mutation in the context of GCC, where the C in the middle is changed to A.

Supplementary Figure 2 A AF38 AF40 HQ21 HQ25 B AF38 AF40 HQ21 HQ25 -1 0 1 -1 0 1 -1 0 1 -1 0 1 AF38 AF40 HQ21 HQ25 B AF38 AF40 HQ21 HQ25

Supplementary Figure 2 Supplementary Figure 2. The T>A mutations in AF-HCC samples. (A) The context of T>A mutations in AF38, AF40, HQ21 and HQ25. The T in middle is the base changed from T to A. (B) The strand bias of T>A mutations in AF-HCC samples. TS: transcribed strand. NTS: non-transcribed strand.

Supplementary Figure 3 A B

Supplementary Figure 3 Supplementary Figure 3. The context of C>A mutations in the whole genome and TP53 mutations. (A) The context of C>A mutations in the whole genome of AF-HCC samples. (B) The top 10 frequent SBS mutations in TP53 from ICGC database.

Supplementary Figure 4 A B C D E F G H I

Supplementary Figure 4 Supplementary Figure 4. Aflatoxin-associated HCC cases in US-HCC and FR-HCC cohorts. (A-C) The three-dimensional scatter plots of the individual mutational pattern in US-HCC, FR-HCC and CN-HCC. The three axes are: the percentage of C>A mutations in all SBSs; the percentage of GCN mutations in all C>A mutations; and the strand bias of C>A mutations. The blue spots are the AF-HCC cases from the high risk region. The black and red spots represent the HCCs in general population in USA (US-HCC), France (FR-HCC) or China (CN-HCC). The red spots (US-AF, FR-AF or CN-AF) are clustered with AF-HCC cases (blue) and depart from the rest US-HCC, FR-HCC or CN-HCC cases (black, US-others, FR-others or CN-others). (D-F) The percentage, context and strand bias of C>A mutations in the 7 US-HCC cases (US-AF) clustered with AF-HCC and the rest US-HCC cases (US-others). (G-I) The percentage, context and strand bias of C>A mutations in the four FR-HCC cases (FR-AF) clustered with AF-HCC and the rest FR-HCC cases (FR-others).

Supplementary Figure 5 A CN-HCC B C FR-HCC US-HCC D E JP1-HCC JP2-HCC TS/NTS (C>A) % GCN % C>A CN-HCC B C TS/NTS (C>A) TS/NTS (C>A) % GCN % GCN % C>A % C>A FR-HCC US-HCC D E TS/NTS (C>A) TS/NTS (C>A) % GCN % GCN % C>A % C>A JP1-HCC JP2-HCC

Supplementary Figure 5 Supplementary Figure 5. Partial aflatoxin-associated HCC cases in general CN-HCC, US-HCC, FR-HCC, JP1-HCC and JP2-HCC cohorts. (A-E) The three-dimensional scatter plots of the individual mutational pattern in HCCs in general population in China (CN-HCC), USA (US-HCC), France (FR-HCC), Japan (JP1-HCC and JP2-HCC). The three axes are: the percentage of C>A mutations in all SBSs; the percentage of GCN mutations in all C>A mutations; and the strand bias of C>A mutations. The blue spots are the AF-HCC cases with typical aflatoxin associated signature. The black, green and red spots represent CN-HCC, US-HCC, FR-HCC, JP1-HCC and JP2-HCC cases. The red spots are clustered with AF-HCC cases (blue) and depart from the rest CN-HCC, US-HCC, FR-HCC, JP1-HCC or JP2-HCC cases (black). The green spots are partially similar to AF-HCC cases (blue) and depart from the rest CN-HCC, US-HCC, FR-HCC, JP1-HCC or JP2-HCC cases (black).

Supplementary Figure 6 A B -1 0 1 -1 0 1 CC-A3M9 DO50821

Supplementary Figure 6 Supplementary Figure 6. Partial aflatoxin-associated HCC cases in US-HCC and FR-HCC cohorts. (A) Distribution of the six mutation classes in the 6 individual cases harbored partial aflatoxin signature in US-HCC or FR-HCC. (B) Sequence contexts of C>A mutations of 2 individual cases harbored partial aflatoxin signature in US-HCC or FR-HCC.

Supplementary Figure 7 A B Distance

Supplementary Figure 7 Supplementary Figure 7. Mutational signature analysis of 49 AF-HCC genomes using using the Wellcome Trust Sanger Institute mutational signatures framework. (A) Identifying the number of processes operative in the 49 AF-HCC samples based on reproducibility of their signatures and low error for re-constructing obtained for K = 1 to 15 signatures. (B) Unsupervised hierarchical clustering of 3 mutational signatures identified in our series (Sig A Sig B and Sig C) and 30 mutational signatures previously identified in a pan-cancer study (Sig 1-30).

Supplementary Figure 8 A B

Supplementary Figure 8 Supplementary Figure 8. Comparison of signature A (sig A) with previously identified signature 24. (A) Signature A and signature 24 are displayed and compared according to the 96 substitution classification. (B) The distance of unsupervised hierarchical clustering between signature A, signature 24 and signature 4, signature 29 previously identified in a pan-cancer study.

Supplementary Figure 9 A

Supplementary Figure 9 Supplementary Figure 9. ADGRB1 mRNA expression level across 84 human tissues. (A) The relative mRNA expression level of ADGRB1 across 84 human tissues. The order of tissue in y axis was ranked by ADGRB1 mRNA expression level from high to low.

Supplementary Figure 10 A B C

Supplementary Figure 10 Supplementary Figure 10. HBV infection of integration rate in AF-HCC detected from whole genome sequencing data. (A) HBV infection rate in blood, normal liver tissue, liver cancer of AF-HCC patients detected from whole genome sequencing data. (B) HBV integration rate in blood, normal liver tissue, liver cancer of AF-HCC patients detected from whole genome sequencing data. (C) The distance of HBV integration break point from TERT gene promoter.

Supplementary Figure 11 A B Density Proportion of signature 24 C

Supplementary Figure 11 Supplementary Figure 11. Mutational signature analysis of 49 AF-HCC and 1072 TCGA/ICGC-HCC genomes using using the Wellcome Trust Sanger Institute mutational signatures framework. (A) Identifying the number of processes operative in the 49 AF-HCC and 1072 TCGA/ICGC-HCC samples based on reproducibility of their signatures and low error for re-constructing obtained for K = 1 to 15 signatures. (B) Distribution of the proportion of signature 24 in each HCC data. (C) Aflatoxin-associated HCC cases identified from 1072 HCCs of general population without known aflatoxin exposure. The aflatoxin-associated HCC cases were identified using the cutoff: proportion of signature 24 > 45% and the total number of SBSs in each tumor > 70.

Supplementary Figure 12 A B

Supplementary Figure 12 Supplementary Figure 12. The correlation between the proportion of 96 mutational classes in urine AFM1 detected samples and urine AFM1 not detected samples. (A-B) The scatter plot and correlation shows the proportion of 96 mutational classes between urine AFM1 detected samples and urine AFM1 not detected samples in AF-HCC. Large scale (A) and small scale (B). R = 0.9780, P < 0.0001, Pearson Correlation.