deOverlapper

deOverlapper is a software tool and algorithm designed to identify and remove overlapping elements within collections of intervals, events, or regions. Its goal is to produce a non-overlapping representation of the data while preserving as much information as possible about the original set.

Inputs commonly include time-based events, spatial regions, or genomic intervals expressed as start and end coordinates. The tool analyzes the overlaps among these items and outputs either a non-overlapping subset, a partition into non-overlapping segments, or a reconstructed sequence with disambiguated boundaries. Metadata may be attached to indicate which original items contributed to each segment.

DeOverlapper can be implemented using several approaches. A graph-based formulation builds an interval graph where overlaps correspond to edges, then uses optimization techniques to select a non-overlapping subcollection or to cover overlaps with splits. Greedy strategies, such as selecting intervals with earliest end times, offer fast, scalable results. Exact methods (dynamic programming or integer linear programming) guarantee optimality for small to moderate data sizes, while heuristic methods are used for large datasets.

Common applications include data cleaning and deduplication, scheduling and resource allocation, multimedia editing to separate overlapping audio or video segments, time-series fusion, and genomic interval processing where precise non-overlapping annotations are required.

Limitations include potential information loss when overlaps are resolved, possible fragmentation of boundaries, and computational complexity on very large datasets. The choice of rule-set—e.g., priority among original items or preferred non-overlapping density—significantly shapes the output. See also interval scheduling, interval graphs, and deduplication.