Xiran Liu Computational biology and population genetics

Projects

Clustering alignment for population structure analysis

Clustering alignment for population structure analysis

Developed methods and tools for aligning latent ancestries across repeated population structure analyses so results remain interpretable across runs and choices of K.

  • Created Clumppling for cluster alignment, mode consolidation, and visualization of correspondence across inferred population structures.
  • Built workflow support for end-to-end population structure analysis and downstream interactive visualization.

Status: published

Clumppling KAlignedoscope Workflow tutorial

Related publications

  • KAlignedoscope: an interactive visualization tool for aligned clustering results
  • Clumppling 2.0: A clustering-alignment program for population-structure analyses
  • Clumppling: cluster matching and permutation program with integer linear programming
  • A Dirichlet model of alignment cost in mixed-membership unsupervised clustering
Alignment-aware single-cell clustering analysis

Alignment-aware single-cell clustering analysis

Extending clustering alignment ideas to the single-cell analysis pipeline to compare models, identify cluster-informative genes, and stabilize downstream interpretation.

  • Developing measures of clustering consistency for robust model comparison.
  • Designing feature-level interpretation metrics to surface informative genes.

Status: published

ML-based methods for multivariate genetic association analysis

ML-based methods for multivariate genetic association analysis

Designed ML-MAGES, a machine-learning framework for shrinkage, pattern discovery, and gene-level aggregation in high-dimensional genetic association studies.

  • Uses neural-network-based supervised learning to account for LD-induced inflation.
  • Combines effect size shrinkage with flexible clustering of multi-trait association patterns.

Status: published

ML-MAGES repository

Related publications

  • ML-MAGES enables multivariate genetic association analyses with genes and effect size shrinkage
  • ML-MAGES: A Machine Learning Framework for Multivariate Genetic Association Analyses with Genes and Effect Size Shrinkage
Mathematical properties of allele-sharing dissimilarities

Mathematical properties of allele-sharing dissimilarities

Analyzing within- and between-population dissimilarity measures as functions of allele-frequency distributions to clarify when and how these statistics should be used.

  • Derives mathematical constraints and interpretable bounds.
  • Connects theory to empirical population-genetic settings.

Status: published

using2025 combinatorics2024 when2023
Cultural evolution through population-genetic methods

Cultural evolution through population-genetic methods

Adapts ideas from genetic evolution to study how cultural traits, social learning, and inherited conventions vary across groups and over time.

  • Uses network-based clustering to recover hierarchical structure in cultural variation.
  • Studies cultural transmission processes such as surnaming decisions and intergenerational inheritance.

Status: published

extracting2022 effects2021