Chromosomal Structural Abnormality Diagnosis by Homologous Similarity

Pathogenic chromosome abnormalities are very common among the general\npopulation. While numerical chromosome abnormalities can be quickly and\nprecisely detected, structural chromosome abnormalities are far more complex\nand typically require considerable efforts by human experts for identification.\nThis paper focuses on investigating the modeling of chromosome features and the\nidentification of chromosomes with structural abnormalities. Most existing\ndata-driven methods concentrate on a single chromosome and consider each\nchromosome independently, overlooking the crucial aspect of homologous\nchromosomes. In normal cases, homologous chromosomes share identical\nstructures, with the exception that one of them is abnormal. Therefore, we\npropose an adaptive method to align homologous chromosomes and diagnose\nstructural abnormalities through homologous similarity. Inspired by the process\nof human expert diagnosis, we incorporate information from multiple pairs of\nhomologous chromosomes simultaneously, aiming to reduce noise disturbance and\nimprove prediction performance. Extensive experiments on real-world datasets\nvalidate the effectiveness of our model compared to baselines.\n