Microdeletions and gross deletions are important causes (~20%) of human inherited disease. Their genomic locations are strongly influenced by the local DNA sequence environment. Yet no systematic study has examined the generative mechanisms. Here, we obtained 42,098 pathogenic microdeletions and gross deletions from the Human Gene Mutation Database (HGMD) that together form a continuum of germline deletions ranging in size from 1 bp to 28,394,429 bp. We analyzed the sequence within 1-kb of the breakpoint junctions and found the frequencies of non-B DNA-forming repeats, GC content, and the presence of seven of 78 specific sequence motifs in the vicinity of pathogenic deletions correlated with deletion length for deletions of length ≤30 bp. Furthermore, we found the repeats of DR, GQ, and STR appear to be important for the formation of longer deletions (>30 bp) but not for the formation of shorter deletions (≤30 bp) and significantly (Chi-square test P-value < 2E-16) more microhomologies were identified in flanking short deletions than long deletions (length >30 bp). We provide evidence to support a functional distinction between microdeletions and gross deletions. A deletion length cut-off of 25-30 bp may serve as an objective means to functionally distinguish microdeletions from gross deletions.

Molecular typing results are not routinely used for CHD. Aiming to uncover the underlying molecular features for different types of CHD, we screened the differentially expressed genes (DEGs) associated with CHD based on the GEO data and expanded in the NCBI-gene and OMIM databases to finally obtain 2021 DEGs. Then the weighted gene co-expression analysis (WGCNA) was performed on the candidate genes, and 6 distinctive WGCNA modules were identified, 2 of which were associated with CHD. Moreover, DEGs were mined as key genes for co-expression based on the module network relationship. Furthermore, the differential miRNAs of CHD and interactions in the database were mined in the GEO dataset to build a multi-factor regulatory network of key genes for co-expression. Combined these results, the CHD samples were further classified into 5 clusters by using the module core genes, and we defined FTH1, HCAR3, RGS2, S100A9 and TYROBP 5 genes in different subgroups. Finally, FTH1, S100A9 and TYROBP were significantly increased while HCAR3 decreased in mRNA level in CHD patients’ blood where RGS2 was hardly determined. In conclusion, the screened core clusters of genes may be potential target for diagnosis and treatment of CHD as a molecular typing module.