Abstract:Abstract Subpopulations of tumor cells characterized by mutation profiles may confer differential fitness and consequently influence the prognosis of cancers. Understanding subclonal architecture has the potential to provide biological insight in tumor evolution and advance cancer precision treatment. Recent subclonal reconstruction methods require heavy computational resources, prior knowledge of the number of subclones, and extensive postprocessing. These drawbacks can be addressed by using a regularized likelihood modeling approach, which is novel to the field. Therefore, we propose a model-based method, Clonal structure identification through pair-wise Penalization, or CliP, to address these drawbacks. To evaluate the performance of CliP against other methods, we generated a benchmark dataset of 4,050 simulated samples with varied tumor purity, read depth, copy number alteration rate, and true numbers of clusters. Our results suggest that CliP outperforms popular methods such as PhyloWGS in accuracy and shows similar robustness in most scenarios. We further compared CliP performance against 10 other subclonal reconstruction methods to the consensus subclonal reconstruction results on whole-genome sequencing (WGS) data from the Pan-Cancer Analysis of Whole Genomes (PCAWG, n = 1,993). Our result shows that of all 11 methods, CliP achieves the highest correlation with the consensus calls. In terms of speed, CliP can finish running a sample with 5,000 SNVs within one minute, which is ~1,000 times faster than MCMC-based algorithms. Next, we profiled the subclonal structures of 7,711 patient samples with well-annotated clinical outcomes from The Cancer Genome Atlas (TCGA) across 32 cancer types applying CliP to the whole-exome sequencing (WES) data. This is the largest and most complete pan-cancer characterization of intratumor heterogeneity (ITH) through the lens of subclonal reconstruction. We further used CliP outputs to address a commonly asked question on which sequencing platform to use for cancer evolution studies: the cost-effect WES data at higher read depth versus the more comprehensive WGS data at lower read depth. There were a total of 588 tumor samples from 21 cancer types, for which both PCAWG and TCGA have profiled using the WGS and WES platform, respectively. Using both datasets to compare results from these samples as benchmark, we observed that for most cancer types, subclonal reconstruction from WES is as informative as that from the matched WGS data. In summary, our study represents a significant methodological advancement in subclonal reconstruction and highlights the importance of measuring tumor subclone structure. Citation Format: Yujie Jiang, Kaixian Yu, Matthew D. Montierth, Shuangxi Ji, Seung Jun Shin, Shuai Guo, Shaolong Cao, Yuxin Tang, Scott Kopetz, Pavlos Msaouel, Jennifer R. Wang, Marek Kimmel, Peter Van Loo, Hongtu Zhu, Wenyi Wang. Pan-cancer analysis of intra-tumor heterogeneity in 9,116 cancers using a novel regularized likelihood model. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4272.

A Novel Multiscale Framework for Delineating Cancer Evolution from Subclonal Compositions.

Simulating the Dynamic Intra-Tumor Heterogeneity and Therapeutic Responses

Effective Sub-clonal Cancer Representation to Predict Tumor Evolution

A Model of Selective Advantage for the Efficient Inference of Cancer Clonal Evolution

A unified computational model for revealing and predicting subtle subtypes of cancers

Reconstructing tumor clonal heterogeneity and evolutionary relationships based on tumor DNA sequencing data

Tumor subclones, where are you?

Evaluation of simulation methods for tumor subclonal reconstruction

Abstract 6931: Characterization of cancer evolution landscape based on accurate detection of somatic mutations in single tumor cells

Integrative Clonal Evolution Analysis on SNV and CNV Levels in Multifocal Breast Cancer

Inferring clonal evolution of tumors from single nucleotide somatic mutations

Heterogeneous Evolution of Breast Cancer Cells—An Endogenous Molecular-Cellular Network Study

Data-Driven Discovery of Mathematical and Physical Relations in Oncology Data Using Human-Understandable Machine Learning

A Bayesian framework to study tumor subclone-specific expression by combining bulk DNA and single-cell RNA sequencing data

Improving personalized prediction of cancer prognoses with clonal evolution models

Abstract 4272: Pan-cancer analysis of intra-tumor heterogeneity in 9,116 cancers using a novel regularized likelihood model

Modeling colorectal cancer evolution

SiCloneFit: Bayesian inference of population structure, genotype, and phylogeny of tumor clones from single-cell genome sequencing data

A Contrastive-Learning-Based Deep Neural Network for Cancer Subtyping by Integrating Multi-Omics Data

[A method for the study of temperature regulation in nonruminant farm animals].

Multi-Omics Data Fusion for Cancer Molecular Subtyping Using Sparse Canonical Correlation Analysis