Supplementary Figure 2: Pipeline of defining B-ALL subtypes.

Details of cohort distribution across different age groups are shown at the top. Gene rearrangements identified by transcriptome sequencing (RNA-seq) were used to define 10 B-ALL subtypes. Available karyotypic information and chromosomal level copy number alterations called from RNA-seq were used to distinguish aneuploidy: high hyperdiploid, low hypodiploid and near haploid. B-ALL subtypes (N = 11, yellow box) showing distinct gene expression profiles and with sufficient number of cases (n ≥ 10) were used as training dataset for Prediction Analysis for Microarray (PAM; Tibshirani, R. et al., Proc Natl Acad Sci U S A. 99, 6567–6572, 2002) to predict “-like” subtypes. Detailed rules are provided in Table 1. chr no., chromosome number.