Fig. 5.

Distribution and multivariate analyses of distinctive molecular features. Boxplots showing the distributions of TIL factor F9 weights (a), PD-L1 gene expression (b), TGF-β expression signature score (c) and TGFB1 gene expression (d) across molecular subtypes of SMC and TCGA. TPM transcript per million. pvalue was determined by Student’s t test: ***p < 0.001; **p < 0.01; *p < 0.05; NS: p ≥ 0.05. The box is bounded by the first and third quartile with a horizontal line at the median and whiskers extend to 1.5 times IQR. e Heatmap showing the statistical significance of associations (−log₁₀p) between distinctive molecular features (rows) and key clinical features (columns) in the combined cohort. Molecular features were ranked by variable usage frequencies in descending order. Clinical features include two continuous variables—Patient age and Tumor purity and five discrete variables—Cohort status: SMC or TCGA; Subtype: ER positive or ER negative; Tumor stage: early (1−2) or late (3−4); Menopausal status: pre or post; Histology subtype: Lobular carcinoma or Invasive ductal carcinoma. Molecular features include S3: HRD-related mutation signature 3; F7: NMF factor 7 associated with ER+ subtype; F9: NMF factor 9 associated with TILs; ESR1-Exp: ESR1 gene expression; HER2-Amp: ERBB2 amplification status; BRCA-Mut: BRCA1/BRCA2 germline pathogenic mutation status; TP53-Mut: TP53 somatic mutation status; TGFB-Sig: TGF-β pathway expression signature