Prediction of lncRNA–Disease Associations via Closest Node Weight Graphs of the Spatial Neighborhood Based on the Edge Attention Graph Convolutional Network

Accumulated evidence of biological clinical trials has shown that long non-coding RNAs (lncRNAs) are closely related to the occurrence and development of various complex human diseases. Research works on lncRNA–disease relations will benefit to further understand the pathogenesis of human complex diseases at the molecular level, but only a small proportion of lncRNA–disease associations has been confirmed. Considering the high cost of biological experiments, exploring potential lncRNA–disease associations with computational approaches has become very urgent. In this study, a model based on closest node weight graph of the spatial neighborhood (CNWGSN) and edge attention graph convolutional network (EAGCN), LDA-EAGCN, was developed to uncover potential lncRNA–disease associations by integrating disease semantic similarity, lncRNA functional similarity, and known lncRNA–disease associations. Inspired by the great success of the EAGCN method on the chemical molecule property recognition problem, the prediction of lncRNA–disease associations could be regarded as a component recognition problem of lncRNA–disease characteristic graphs. The CNWGSN features of lncRNA–disease associations combined with known lncRNA–disease associations were introduced to train EAGCN, and correlation scores of input data were predicted with EAGCN for judging whether the input lncRNAs would be associated with the input diseases. LDA-EAGCN achieved a reliable AUC value of 0.9853 in the ten-fold cross-over experiments, which was the highest among five state-of-the-art models. Furthermore, case studies of renal cancer, laryngeal carcinoma, and liver cancer were implemented, and most of the top-ranking lncRNA–disease associations have been proven by recently published experimental literature works. It can be seen that LDA-EAGCN is an effective model for predicting potential lncRNA–disease associations. Its source code and experimental data are available at https://github.com/HGDKMF/LDA-EAGCN.

Influences of rare copy number variation on human complex traits

The human genome contains hundreds of thousands of regions exhibiting copy number variation (CNV). However, the phenotypic effects of most such polymorphisms are unknown because only larger CNVs (spanning tens of kilobases) have been ascertainable from the SNP-array data generated by large biobanks. We developed a new computational approach that leverages abundant haplotype-sharing in biobank cohorts to more sensitively detect CNVs co-inherited within extended SNP haplotypes. Applied to UK Biobank, this approach achieved 6-fold increased CNV detection sensitivity compared to previous analyses, accounting for approximately half of all rare gene inactivation events produced by genomic structural variation. This extensive CNV call set enabled the most comprehensive analysis to date of associations between CNVs and 56 quantitative traits, identifying 269 independent associations (P < 5 x 10-8) - involving 97 loci - that rigorous statistical fine-mapping analyses indicated were likely to be causally driven by CNVs. Putative target genes were identifiable for nearly half of the loci, enabling new insights into dosage-sensitivity of these genes and implicating several novel gene-trait relationships. CNVs at several loci created extended allelic series including deletions or duplications of distal enhancers that associated with much stronger phenotypic effects than SNPs within these regulatory elements. These results demonstrate the ability of haplotype-informed analysis to empower structural variant detection and provide insights into the genetic basis of human complex traits.

Fish and chips: the origin of human gene families is a predictor of the location of GWAS signals

GWAS have identified thousands of loci associated with human complex diseases and traits. How these loci are distributed through the genome has not been systematically evaluated. We hypothesised that the location of GWAS loci differ between ancestral linkage groups (ALGs) related to the paralogy and function of genes. We used data from the NHGRI-EBI GWAS catalog to determine whether the density of GWAS loci relative to HapMap variants in each ALG differed, and whether ALG’s were enriched for experimental factor ontological (EFO) terms assigned to the GWAS traits. In a gene-level analyses we explored the characteristics of genes linked to GWAS loci and those mapping to the ALG’s. We find that GWAS loci were enriched or deficient in 9 and 7 of the 17 ALG’s respectively, while there was no difference in the number of GWAS loci in regions of the human genome unassigned to an ALG. All but 2 ALG’s were significantly enriched or deficient for one or more EFO terms. Lastly, we find that genes assigned to an ALG are under higher levels of selective constraint, have longer coding sequences and higher median expression in the tissue of highest expression than genes not mapping to an ALG. On the other hand, genes associated with GWAS loci have longer genomic length and exhibit higher levels of selective constraint relative to non-GWAS genes.Collectively, this suggests that understanding the location and ancestral origins of GWAS signals may be informative for the development of tools for variant prioritization and interpretation.

Highly pleiotropic variants of human traits are enriched in genomic regions with strong background selection

Recent studies have shown the ubiquity of pleiotropy for variants affecting human complex traits. These studies also show that rare variants tend to be less pleiotropic than common ones, suggesting that purifying natural selection acts against highly pleiotropic variants of large effect. Here, we investigate the mean frequency, effect size and recombination rate associated with pleiotropic variants, and focus particularly on whether highly pleiotropic variants are enriched in regions with putative strong background selection. We evaluate variants for 41 human traits using data from the NHGRI-EBI GWAS Catalog, as well as data from other three studies. Our results show that variants involving a higher degree of pleiotropy tend to be more common, have larger mean effect sizes, and contribute more to heritability than variants with a lower degree of pleiotropy. This is consistent with the fact that variants of large effect and frequency are more likely detected by GWAS. Using data from four different studies, we also show that more pleiotropic variants are enriched in genome regions with stronger background selection than less pleiotropic variants, suggesting that highly pleiotropic variants are subjected to strong purifying selection. From the above results, we hypothesized that a number of highly pleiotropic variants of low effect/frequency may pass undetected by GWAS.