AIEdit manuscript published: alignment-free genome assembly polishing using spaced seed patterns
news ·We are pleased to announce that our scientific manuscript, AIEdit: Alignment-free genome assembly polisher trained on spaced seed match patterns, is now published in PLOS Computational Biology.
AIEdit is a machine learning-based genome assembly polisher designed to operate without sequence alignments while remaining scalable across sequencing platforms. By combining spaced seed matching with a neural network trained to detect and correct dense sequencing error patterns, AIEdit performs accurate base-level correction in an alignment-free framework.
Across both simulated and experimental long-read datasets, AIEdit demonstrates strong accuracy and computational efficiency. On simulated human assemblies with high error rates, AIEdit reduced errors by 58% compared to ntEdit’s 21%, while running substantially faster than alignment-based polishing approaches. On experimental Oxford Nanopore human genome assemblies, AIEdit improved the Merqury quality score from QV 28.7 to 32.9, outperforming existing k-mer-based polishers and achieving comparable accuracy to Medaka in a fraction of the runtime.