Lambuta, R.-A. | Search bioRxiv

bioRxiv2026

16p11.2 Copy Number Variation Alters Genome Architecture and Transcriptional Regulation During Neurodevelopment

Microdeletions and microduplications in 16p11.2 are responsible for a spectrum of neurodevelopmental disorders (NDDs) with partially reciprocal and overlapping symptoms. However, the clinical variability in 16p11.2 microduplication patients is significantly greater than that in patients carrying a microdeletion. Here, we use iPSCs derived from members of a family carrying a 16p11.2 microduplication and model neurodevelopment through in vitro differentiation into neural progenitor cells (NPCs) and neurons. The analysis is complemented by reanalysis of publicly available data from 16p11.2 microdeletion patients. Transcriptome analysis revealed MAPK3-centered hubs of upregulated genes in the microduplication-carrying and downregulated in the microdeletion-carrying cells, indicating that MAPK3 is a central driver of 16p11.2 Copy Number Variation (CNV) pathology. While genes within the 16p11.2 region showed about a two-fold reduction in expression in cells carrying the microdeletion, their expression levels in microduplication-derived NPCs and neurons, but not in iPSCs, were elevated to a degree much higher than expected. This observation was accompanied by a substantial number of dysregulated genes unrelated to the genes in the critical region or their interaction networks. To further investigate whether altered chromatin organization may accompany these transcriptional changes, we generated Hi-C data from patient and control iPSCs and NPCs. This revealed increased chromatin contacts within the duplicated 16p11.2 region in patient-derived iPSCs, while genome-wide compartment analysis showed that increased compartments preferentially co-occurred with upregulated differentially expressed genes, particularly in NPCs. Together, these findings suggest that 16p11.2 microduplication may influence gene expression through both local dosage-dependent mechanisms and broader, differentiation-associated changes in chromatin organization. Our data support a model in which CNV-associated genome architecture changes may modulate transcriptional dysregulation and contribute to the variable neurodevelopmental phenotypes associated with 16p11.2 rearrangements.

pathology↗

bioRxiv2026

Correcting Preprocessing Bias in Sparse Chromatin Contact Data Enables Physically Interpretable Reconstruction of Genome Architecture

DNA is the largest biopolymer in nature, and chromatin contact maps are widely interpreted as quantitative readouts of its three-dimensional organization. However, the validity of such interpretations critically depends on how these maps are processed. Here, we identify a previously overlooked but fundamental source of bias in chromatin contact data analysis. We demonstrate that a widely adopted preprocessing convention, namely whole-matrix percentile clipping, systematically distorts sparse contact maps by collapsing their dynamic range. This effect is strongest in near-diagonal interactions, precisely the regime encoding chromatin domains and looping structures, thereby compromising quantitative interpretation while preserving superficial structural features. We show that this distortion represents a sparsity-dependent failure mode of current preprocessing standards and affects the comparability of datasets and computational methods across technologies and sequencing depths. To address this, we introduce a statistically consistent preprocessing framework based on nonzero-percentile clipping and log-space normalization, which preserves the intrinsic dynamic range of observed contacts. Building on this foundation, we present CCUT, a modular deep learning framework for chromatin contact map reconstruction. Under corrected preprocessing, reconstructed maps recover domain organization, contact decay, and scaling behavior consistent with polymer physics. Importantly, we demonstrate quantitative agreement between reconstructed maps and simulated contact patterns derived from a kinetic Monte Carlo loop extrusion model, enabling direct comparison between experimental data and physical models. Together, our results establish preprocessing as a decisive determinant of the physical interpretability of chromatin contact maps and provide a principled framework for robust and comparable analysis across chromatin conformation capture technologies.

biophysics↗