MapSplice is a second-generation algorithm designed for accurate detection of splice junctions in RNA-seq data. It focuses on high sensitivity and specificity, CPU and memory efficiency, and the ability to detect both canonical and non-canonical splices. MapSplice can handle both short ( approximately 75 bp) and long reads (≥75 bp). The algorithm leverages the quality and diversity of read alignments to increase accuracy. Synthetic data and experimental validation using breast cancer RNA-seq datasets demonstrate that MapSplice outperforms TopHat and SpliceMap in terms of sensitivity and specificity. MapSplice accurately identifies alternative splicing events, including exon skipping, alternative 3′-end, alternative 5′-start, and mutually exclusive exons, with high concordance to previous studies. The algorithm also reveals significant differences in splicing patterns between molecular subtypes of breast cancer, providing insights into the extensive nature of alternative splicing.MapSplice is a second-generation algorithm designed for accurate detection of splice junctions in RNA-seq data. It focuses on high sensitivity and specificity, CPU and memory efficiency, and the ability to detect both canonical and non-canonical splices. MapSplice can handle both short ( approximately 75 bp) and long reads (≥75 bp). The algorithm leverages the quality and diversity of read alignments to increase accuracy. Synthetic data and experimental validation using breast cancer RNA-seq datasets demonstrate that MapSplice outperforms TopHat and SpliceMap in terms of sensitivity and specificity. MapSplice accurately identifies alternative splicing events, including exon skipping, alternative 3′-end, alternative 5′-start, and mutually exclusive exons, with high concordance to previous studies. The algorithm also reveals significant differences in splicing patterns between molecular subtypes of breast cancer, providing insights into the extensive nature of alternative splicing.