Search

Additional file 3: Figure S3. RNA-seq read coverage of CD44 exon 5 to exon 16 for all samples in CRC18P and CRC10P datasets. Exons v8–10 showed more inclusion in CRC and MC samples (labeled in red).

Additional file 12: Figure S12. Splicing events identified in this study confirmed by TCGA junction expression data. Boxplots of junction usage of 16 junctions of 9 genes in 51 normal tissue and 382 CRC or metastatic tissue (tumor). P-values are based on Wilcoxon Rank-Sum Test. Nominal P-values are shown and adjusted P-values using Bonferroni correction can be found in the supplementary Table 7. Dots in the boxplot represent individual patient in TCGA.

Additional file 11: Figure S11. Boxplot of PSI values for genes in Figs. 4 and 5. Only AS events identified using PSI methods were plotted here.

Additional file 8: Figure S8. RNA-seq read coverage of SERPINA1 exon 1 to exon 4 in CRC18P dataset. Exon E1a showed more exclusion/downregulation in MC samples (labeled in blue).

Additional file 5: Figure S5. RNA-seq read coverage of HKDC1 alternative first exons in CRC18P and CRC10P datasets. Exons E1a, E2a and E3a showed more inclusion/upregulation in CRC and MC samples (labeled in red).

Additional file 9: Figure S9. RNA-seq read coverage of CALD1 exon 5 to exon 7 in CRC18P dataset. Exons E5b and E6 showed more exclusion/downregulation in MC samples (labeled in blue).

Additional file 6: Figure S6. RNA-seq read coverage of CHEK1 alternative first exons in CRC18P and CRC10P datasets. Exon E1a showed more exclusion/downregulation in CRC and MC samples (labeled in blue).

Additional file 7: Figure S7. RNA-seq read coverage of HNF4A alternative first exons in CRC18P and CRC10P datasets. Exon 1A showed more exclusion/downregulation in CRC and MC samples (labeled in blue).

Additional file 2: Figure S2. RNA-seq read coverage of COL6A3 exon 6 and 7 (E6, E7) for all samples in CRC18P and CRC10P datasets. E6 showed more inclusion in CRC and MC samples (labeled in red).

Additional file 8: Figure S8. RNA-seq read coverage of SERPINA1 exon 1 to exon 4 in CRC18P dataset. Exon E1a showed more exclusion/downregulation in MC samples (labeled in blue).

Additional file 4: Figure S4. RNA-seq read coverage of ARHGEF9 alternative first exons in CRC18P and CRC10P datasets. Exon E1a showed more exclusion/downregulation in CRC and MC samples (labeled in blue).

Additional file 6: Figure S6. RNA-seq read coverage of CHEK1 alternative first exons in CRC18P and CRC10P datasets. Exon E1a showed more exclusion/downregulation in CRC and MC samples (labeled in blue).

Additional file 7: Figure S7. RNA-seq read coverage of HNF4A alternative first exons in CRC18P and CRC10P datasets. Exon 1A showed more exclusion/downregulation in CRC and MC samples (labeled in blue).

Additional file 9: Figure S9. RNA-seq read coverage of CALD1 exon 5 to exon 7 in CRC18P dataset. Exons E5b and E6 showed more exclusion/downregulation in MC samples (labeled in blue).

Additional file 1: Figure S1. RNA-seq read coverage of CLK1 exon 4 and 5 (E4, E5) for all samples in CRC18P and CRC10P datasets. E4 showed more inclusion in CRC and MC samples (labeled in red).

Additional file 5: Figure S5. RNA-seq read coverage of HKDC1 alternative first exons in CRC18P and CRC10P datasets. Exons E1a, E2a and E3a showed more inclusion/upregulation in CRC and MC samples (labeled in red).

Additional file 13: Figure S13. Logistic regression predicts sample type using junction usage data. (A) Spearman’s correlation coefficient of the junction usages of the TCGA data for the 8 genes. The 8 genes were separated to 3 groups (as indicated) based on the correlation values. (B) AUC scores of the ROC curves for the logistic regression using different number and combinations of the predictors. Average value of the five AUC scores for each formula were shown and the error bar is the standard deviation. Each AUC score was generated using randomly selected half of the data to train a logistic regression model and tested on the rest of the data.