cue: A deep learning framework for cross-platform SV discovery and genotyping. It translates sequence alignments into multi-channel images that encode SV-informative signals by juxtaposing distinct genome intervals, and uses a stacked hourglass convolutional network to detect SV breakpoints stratified by type and genotype.

cue2: A deep learning framework for SV discovery (extension of cue) optimized for long reads (e.g., PacBio HiFi).

ralphi: Deep reinforcement learning framework for haplotype assembly. It combines a graph convolutional network with an actor-critic model to accurately assign read fragments to their respective haplotypes under the maximum fragment cut (MFC) objective.

marti: Lightweight framework for classification and quantification of artifactual long-read cDNA constructs. It detects and annotates PCR, RT, and sequencing artifacts in PacBio and ONT cDNA libraries.

insilicoSV: Grammar-based framework for SV simulation and placement. It models SVs using a simple and flexible grammar (e.g., ABC → aBBBc), allowing users to define standard and custom genome rearrangements, as well as encode genome placement constraints. It includes a WDL pipeline with support for genome simulation (including genome mixtures), read simulation, alignment, and visualization.

VARium: An extensive suite of > 100 synthetic genomes designed to systematically assess the performance of structural variant (SV) discovery tools as a function of key domain parameters and confounders. It provides multi-platform, multi-coverage simulated WGS datasets with SVs stratified by type, size, genomic context, and genome complexity.

verix: Toolkit for benchmarking and harmonization of complex structural variants.

svvs: Grammar-based complex SV signature visualizer and solver. Links rearrangement expressions to their alignment-based genomic signals — split-read signatures, copy number profiles, and breakpoint graphs — and can invert the problem: given an observed signature, it enumerates all grammatical expressions consistent with it.