MSc Thesis Defense: Mekan Myradov, A SYSTEMATIC EVALUATION OF SINGLE-CELL BATCH INTEGRATION METRICS AND sBEE: A ROBUST NEW METRIC, Date & Time: 13 July, 2026 – 11:30 AM, Place: L067

A SYSTEMATIC EVALUATION OF SINGLE-CELL BATCH INTEGRATION METRICS AND sBEE: A ROBUST NEW METRIC

Mekan Myradov
Data Science, MSc Thesis, 2026

Thesis Jury

    Assoc. Prof. Öznur TAŞTAN (Thesis Advisor)

  Asst. Prof. Onur VAROL

  Prof. Tunahan ÇAKIR

Date & Time: July 13th, 2026 –  11:30 AM

Place: FENS L067

Keywords: single-cell RNA sequencing, batch effect correction, batch integration, benchmarking, evaluation metric

Abstract

Single-cell RNA sequencing datasets generated across laboratories and experimental conditions often contain batch effects that obscure biological variation. This thesis systematically evaluates widely used batch integration metrics under controlled synthetic scenarios that isolate common integration challenges, including imbalanced batch composition, partial cell-type overlap, heterogeneous cluster densities, and varying cluster geometries. The analyses reveal that existing metrics capture different aspects of integration quality and can fail under specific data structures, producing discordant rankings of integration methods. Motivated by these findings, this thesis introduces sBEE, the single-cell Batch Effect Evaluator, a metric that quantifies batch mixing within each cell type using complementary cross-batch distance and local neighborhood composition signals. Across simulated scenarios, sBEE provides stable and interpretable evaluations while remaining robust to batch imbalance and partial cell-type overlap. The metric is further evaluated on real single-cell RNA sequencing datasets, where it behaves consistently with the controlled simulations. Together, these results provide a principled framework for interpreting batch integration metrics and support sBEE as a robust metric for evaluating single-cell batch mixing.