JuxtaBAM flags mismatched samples in joint multi-omic datasets

A newly described computational method called JuxtaBAM can identify inconsistent, duplicated or potentially mislabeled samples in datasets that combine epigenomic and transcriptomic measurements. The software works without predefined panels of population-specific single-nucleotide polymorphisms and can be applied across species, according to an accepted study in BMC Genomics. Its central application is data validation: JuxtaBAM compares genetic variation embedded in matched sequence files and statistically flags sample pairs that may not represent the biological samples they are claimed to represent.

The authors developed the method in response to a growing challenge in integrative molecular analysis. Joint profiling of regulatory activity and gene expression can strengthen the statistical basis for identifying disrupted molecular pathways associated with common human disease, but those analyses require correctly identified samples. At the same time, many existing tools for detecting sample swaps depend on predetermined lists of population-based SNPs or linkage-disequilibrium maps. Such dependence can limit their applicability to model organisms and to the particular genomic regions interrogated by a given assay. JuxtaBAM instead bases its validation on genetic variation found directly within the input BAM files, allowing the evidence for sample identity to come from the datasets themselves.

Technically, JuxtaBAM identifies single-nucleotide variants in high-signal regions that overlap in the input epigenomic and transcriptomic BAM files. Concentrating on these shared regions allows the software to focus genotyping on positions with sufficient read coverage in both assays. It then computes pairwise SNV mismatch rates across all samples and models the counts of mismatched genotypes with a beta-binomial distribution. The statistical model provides a quantitative test for sample mislabeling and duplication, while the final output identifies pairs that warrant investigation as possible mismatches or duplicates. This formulation makes the method independent of a fixed population-SNP panel, although its usefulness still requires a reference genome and shared, adequately covered sequence regions from which informative variation can be identified.

The researchers demonstrated JuxtaBAM on a joint dataset generated in their earlier work, comprising paired ATAC-seq and RNA-seq profiles from a panel of 25 gastric cancer cell lines. In that application, the method identified an inconsistency involving a duplicated sample. The result showed that JuxtaBAM could detect a concrete labeling problem within a dataset containing both chromatin-accessibility and transcriptomic measurements. The study also evaluated the method using ENCODE mouse DNase-seq and RNA-seq data from cell lines derived from two different mouse strains. These data included multiple experiments generated by independent laboratories and covered different assay types, providing a test of the software’s operation across laboratory sources, species and profiling formats. Together, the demonstrations support JuxtaBAM’s use as a general validation tool rather than as a source of new biological findings about gastric cancer or mouse cells.

JuxtaBAM’s principal advantage is that it does not require researchers to match a sample set to an external reference panel built around known population variation. Variants are discovered directly from the data being validated, allowing the approach to be considered for any organism with a reference genome when the joint datasets contain overlapping regions with adequate read coverage. The reported validations nevertheless have defined limits. JuxtaBAM produces statistical flags for possible sample mismatch or duplication; it does not by itself establish which member of a flagged pair is mislabeled or explain how the error arose. Its performance also depends on the availability of informative shared variants. Within those constraints, the study presents a species-independent and SNP-panel-independent framework for checking the integrity of joint epigenomic and transcriptomic profiling datasets.

Subject of Research: Software validation of sample identity in joint epigenomic and transcriptomic profiling datasets

Article Title: JuxtaBAM: a species and SNP-independent method to validate the integrity of joint epigenomic and transcriptomic profiling datasets

Article References: Razavi-Mohseni, M., & Beer, M. A. (2026). JuxtaBAM: a species and SNP-independent method to validate the integrity of joint epigenomic and transcriptomic profiling datasets. BMC Genomics. https://doi.org/10.1186/s12864-026-13485-1

Image Credits: AI Generated

DOI: 10.1186/s12864-026-13485-1

Keywords: JuxtaBAM, multi-omics, sample validation, ATAC-seq, RNA-seq, SNV genotyping, beta-binomial model, DNase-seq, gastric cancer cell lines, bioinformatics, species, SNP-independent