Download Supplementary Figure Legends

Supplementary Figure Legends
Figure S1. Somatic mutations of each tumor. Missense, nonsense and splicing identified by
whole-exome sequencing analysis
Figure S2. EML4-ALK fusion gene detected from TH2 and TH5. (A) The break point region is
located within intron 19 of ALK and intron 6 of EML4. (B-D) The IGV visualization of EML4-ALK
fusion in TH1 from RNA-seq data, in TH1 from panel-sequencing data and in TH4 from RNAseq data.
Figure S3. The genomic and transcriptomic differences between multiple lymph nodes of
patients TH2, TH3, TH4 and TH6. The density plots of cancer cell fraction (CCF) between the
primary site and lymph node metastases are shown in upper panels. The scatter plots for
similarity of gene expression profiles between the primary site and lymph node metastases are
shown in lower panels. Differentially expressed cancer related-genes (>2-fold change) are
highlighted in red and green colors.
Figure S4. The genomic and transcriptomic pattern of TH6 case. (A) The mutation
heatmap. (B) The scatter plot shows the similarity of gene expression profiles between nodes
(using 15,842 genes with at least > 5 reads) by measuring Pearson’s correlation coefficient (r).
(C) The density plot of cancer cell fraction (CCF) between the primary and the lymph node. (D)
The gene expression pattern of genes belonging to SQCCs, PI(3)K/RTK/RAS signaling. Gene
expression values in lymph nodes were mapped on the distribution of gene expression in TCGA
SQCCs (n=495), represented by the boxplot. AKT2 and RASA1 show significant clinical
outcome difference in the gene expression of TCGA SQCCs patients; *P < 0.05.
Figure S5. The measurement representing early vs. late metastasis. (A) The percentage of
trunk/non-trunk mutations for six cases. (B) The expression levels of AKT2 gene as a potential
early event in tumorigenesis.