Beyond Discrete Ancestry Groups: Distribution-Aware In-Context Learning for Cancer Outcome Prediction Across the Genetic Ancestry Continuum
Organizations: Department of Genetics, Genomics and Informatics, University of Tennessee Health Science Center, Memphis, TN, USA
Abstract
Machine learning models for omics-based prediction of cancer outcomes are commonly built on aggregated data without distinguishing ancestry groups, or under an ancestry-discrete design, which partitions samples into ancestry groups and fits a separate model within each group. Recent evidence shows that this approach is inherently limited: predictive accuracy declines with genetic distance from the training samples, even within a single ancestry group, because the data distribution shifts continuously rather than only between predefined categories. In this paper, we propose distribution-aware in-context learning (DA-ICL) for ancestry-continuum modeling. Our method extends conventional ancestry-discrete modeling to ancestry-continuum modeling, capturing the nonstationary omics-endpoint relationship across the genetic continuum while enabling efficient modeling through an in-context learning foundation model. We provide theoretical analysis demonstrating the superiority of the ancestry-continuum over ancestry-discrete modeling, along with empirical results supporting the proposed framework. Source code is available at https://github.com/ai4pm/DA-ICL-Continuum.
Figures & tables
| KIRC-OS-mRNA | KIRC-PFI-mRNA | UCEC-OS-mRNA | UCEC-PFI-mRNA | BRCA-OS-mRNA | BRCA-PFI-mRNA | KIRC-OS-miRNA | KIRC-PFI-miRNA | UCEC-OS-miRNA | UCEC-PFI-miRNA | BRCA-OS-miRNA | BRCA-PFI-miRNA | KIRC-OS-Meth | KIRC-PFI-Meth | UCEC-OS-Meth | UCEC-PFI-Meth | BRCA-OS-Meth | BRCA-PFI-Meth | |
| DA-ICL vs EN | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 | – [2pt]0.829 | – [2pt]0.461 | – [2pt]0.059 | [2pt]0.034 | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 | – [2pt]0.746 | – [2pt]0.089 | – [2pt]0.502 | – [2pt]0.400 | ✓ [2pt] 0.001 | – [2pt]0.428 | ✓ [2pt] 0.001 | ✓ [2pt]0.011 |
| DA-ICL vs RF | ✓ [2pt]0.004 | – [2pt]0.088 | [2pt]0.021 | – [2pt]0.187 | [2pt]0.005 | [2pt]0.013 | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 | ✓ [2pt]0.008 | ✓ [2pt] 0.001 | ✓ [2pt]0.036 | – [2pt]0.074 | – [2pt]0.938 | – [2pt]0.495 | ✓ [2pt]0.002 | – [2pt]0.187 | ✓ [2pt]0.001 | ✓ [2pt]0.007 |
| DA-ICL vs XGB | ✓ [2pt]0.001 | – [2pt]0.063 | [2pt]0.026 | – [2pt]0.335 | – [2pt]0.877 | – [2pt]0.332 | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 | ✓ [2pt]0.030 | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 | – [2pt]0.821 | – [2pt]0.328 | ✓ [2pt] 0.001 | – [2pt]0.185 | ✓ [2pt] 0.001 | ✓ [2pt] 0.001 |
| KIRC-OS-mRNA | KIRC-PFI-mRNA | UCEC-OS-mRNA | UCEC-PFI-mRNA | BRCA-OS-mRNA | BRCA-PFI-mRNA | KIRC-OS-miRNA | KIRC-PFI-miRNA | UCEC-OS-miRNA | UCEC-PFI-miRNA | BRCA-OS-miRNA | BRCA-PFI-miRNA | KIRC-OS-Meth | KIRC-PFI-Meth | UCEC-OS-Meth | UCEC-PFI-Meth | BRCA-OS-Meth | BRCA-PFI-Meth | |
| ICL-Discrete | ||||||||||||||||||
| ICL-Continuum | ||||||||||||||||||
| DA-ICL-Cont. |