FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4320078097> ?p ?o ?g. }

• W4320078097 abstract "Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Methods Data availability References Decision letter Author response Article and author information Metrics Abstract Individuals infected with the SARS-CoV-2 virus present with a wide variety of symptoms ranging from asymptomatic to severe and even lethal outcomes. Past research has revealed a genetic haplotype on chromosome 3 that entered the human population via introgression from Neanderthals as the strongest genetic risk factor for the severe response to COVID-19. However, the specific variants along this introgressed haplotype that contribute to this risk and the biological mechanisms that are involved remain unclear. Here, we assess the variants present on the risk haplotype for their likelihood of driving the genetic predisposition to severe COVID-19 outcomes. We do this by first exploring their impact on the regulation of genes involved in COVID-19 infection using a variety of population genetics and functional genomics tools. We then perform a locus-specific massively parallel reporter assay to individually assess the regulatory potential of each allele on the haplotype in a multipotent immune-related cell line. We ultimately reduce the set of over 600 linked genetic variants to identify four introgressed alleles that are strong functional candidates for driving the association between this locus and severe COVID-19. Using reporter assays in the presence/absence of SARS-CoV-2, we find evidence that these variants respond to viral infection. These variants likely drive the locus’ impact on severity by modulating the regulation of two critical chemokine receptor genes: CCR1 and CCR5. These alleles are ideal targets for future functional investigations into the interaction between host genomics and COVID-19 outcomes. Editor's evaluation A genetic haplotype on chromosome 3 that entered the human lineage from mating with Neanderthals has previously been implicated as a strong genetic risk factor for severe COVID-19 outcomes. This study uses population genetics and functional genomics tools along with experimental assays to assess the genetic variants in these regions for their likelihood of driving the severe COVID-19 phenotype. They ultimately identify 4 (out of about 600) variants as strong functional candidates. This study is a valuable contribution to the interaction between host genomics and COVID-19 outcomes and provides compelling evidence allowing for more targeted future functional investigations. https://doi.org/10.7554/eLife.71235.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Since its emergence in late 2019, SARS-CoV-2 has infected more than 160 million people worldwide and claimed more than 3 million lives (WHO, 2021). The variance in patient outcomes is extreme, ranging from no ascertainable symptoms in some cases to fatal respiratory failure in others (Vetter et al., 2020). This wide range of patient outcomes is due in part to comorbidities; however, prior health conditions do not explain the full range of outcomes (Zhou et al., 2020). Therefore, efforts have been made to assess a potential genetic component. Repeatedly, a region on chromosome 3 encompassing a cluster of chemokine receptor genes has been reported as having a strong association with an increase in COVID-19 severity in Europeans, with the strongest reported risk variant conferring an odds ratio of 1.88 for requiring hospitalization (p=2.7*10–49, COVID-19 Host Genetics Initiative, 2021). Zeberg and Pääbo, 2020 identified that the strongest COVID-19 severity locus was introgressed by Neanderthals, with a core introgressed haplotype spanning ~49 kb from chr3:45,859,651–45,909,024 (hg 19) including rs35044562, reported as one of the leading variants of the association (COVID-19 Host Genetics Initiative, 2020) and a broader, extended haplotype with reduced linkage spanning ~333 kb from chr3:45,843,315–46,177,096. Subsequently, this locus was fine-mapped to two independent risk signals, one which is confirmed to fall within the Neanderthal haplotype and tagged by a set of strongly linked SNPs including rs35044562 and rs10490770, while the other, led by rs2271616, falls just upstream (COVID-19 Host Genetics Initiative, 2021; Kousathanas et al., 2022). The core haplotype is at highest frequency in South Asian populations (30%), as well as at appreciable frequency in Europe (8%) and the Americas (4%), yet it is virtually absent in East Asia. The stark difference in frequency between South Asian and East Asian populations implies that the haplotype may have been positively selected in South Asian populations, for which there is support (Racimo et al., 2014; Jagoda et al., 2018; Browning et al., 2018) and/or subject to purifying selection in East Asian populations. However, the specific phenotypic consequences of this haplotype leading to its potential adaptive effect as well as its effect on COVID-19 severity remain unknown. Moreover, the potential causal drivers of the selective pressure, as well as COVID-19 severity remain unstudied. Here, we identify putative functional variants within this haplotype that may be driving its association with COVID-19 severity. To do so, we first examine the haplotype in the context of a broader introgressed segment. We then identify loci within the introgressed segment that are associated with levels of gene expression (eQTLs) in vivo. We next compare the eQTL effects of these variants with differentially expressed genes in COVID-19 and related infection datasets to identify which response genes for these eQTLs are potentially relevant to the COVID-19 phenotype. We follow this computational approach with a high-throughput functional Massively Parallel Reporter Assay (MPRA) and identify 20 variants along the introgressed segment that directly modulate reporter gene expression. We intersect these 20 variants with a host of molecular and phenotypic datasets to further refine them to 4 which display the strongest evidence of contributing to the genetic association with severe COVID-19 at this locus. We then investigated these four variants (eight alleles) using reporter assays in the context of the promoter of their most likely endogenous target gene (CCR1 or CCR5), and in the presence/absence of replicating SARS-CoV-2, revealing evidence of important functionality. These tested variants primarily modulate expression through their potential effects on CCR1 and CCR5 cis-regulation and are strong candidate variants that should be investigated with future targeted functional experiments. An overview of this experimental workflow is shown in Figure 1. Figure 1 Download asset Open asset Overview of experimental workflow from whole genome scans for Neanderthal introgression to variant section for the MPRA and SARS-CoV-2 infection reporter assay experiments. Results Genome-wide scans for Neanderthal introgression We carried out two genome-wide searches for introgressed loci in a European population for which we also had available eQTL data using Sprime (Browning et al., 2018) and U and Q95 (Racimo et al., 2014) methods. We used 423 Estonian whole genome samples (Pankratov et al., 2020) that constitute a well-studied representative sample of the broader Estonian population as sampled by the Estonian Biobank (EGCUT) (Leitsalu et al., 2015). These samples also have available whole blood RNA-sequence data which contributed to eQTLGen, a broad whole blood eQTL analysis study (Võsa et al., 2018). By utilizing genomes that were part of the eQTL study population, we can be assured that the associations between alleles and gene expression is accurate, as differential linkage disequilibrium (LD) between alleles in different populations can decrease the efficacy of using eQTL data from one population on another. We initially conducted the Sprime scan (Browning et al., 2018) using the 423 Estonians as the ingroup population along with 36 African samples from the Simons Genome Diversity Project (SGDP) with no evidence of European admixture (Mallick et al., 2016) as an outgroup (Supplementary file 1a). From this scan, we identified 175,550 likely archaically introgressed alleles across 1,678 segments (Supplementary file 1b). Following Browning et al., 2018, we then identified segments as confidently introgressed from Neanderthals if they had at least 30 putatively archaically introgressed alleles with a match rate to the Vindija Neanderthal genome (Prüfer et al., 2017) greater than 0.6 and a match rate with the Denisovan genome (Meyer et al., 2012) less than 0.4 (Browning et al., 2018). In total, we identified 693 such segments (Supplementary file 1c), including the segment containing the COVID-19 severity haplotype on chromosome 3 (see above). We next used the U and Q95 scan, which specifically identifies regions of introgression showing evidence of positive selection (Racimo et al., 2017). Using Africans from the 1000 genomes project as an outgroup (Auton et al., 2015), we found 493 such regions (Supplementary file 1d). We did not detect the introgressed COVID-19 severity haplotype in our population via this method. This suggests that the COVID-19 severity associated segment, while likely introgressed from Neanderthals based on its detection in our Sprime scan and via the work of others (Zeberg and Pääbo, 2020), was not under positive selection in the Estonian population. This is consistent with the previously reported lower frequency (8%) of the haplotype in Europeans relative to South Asian populations in which the haplotype is at higher frequency (30%) (Zeberg and Pääbo, 2020). However, U and Q95 scans do detect this region in South Asian populations (Racimo et al., 2017; Jagoda et al., 2018), supporting positive selection on this haplotype in South Asian, but not European populations. We next examined which alleles in these putatively Neanderthal introgressed regions detected using these two genome-wide scans also are cis- and trans-eQTLs in the eQTLGen whole blood dataset (Võsa et al., 2018). From the U and Q95 data, we identified 684 cis-eQTLs across 250 40 kb windows (Supplementary file 1e). There were no trans-eQTLs detected in this set. From the Sprime data, we found 27,342 cis-eQTLs from 318 segments along with four trans-eQTLs from three segments (Supplementary file 1f and Supplementary file 1g). Refinement of the severe COVID-19 associated introgressed segment In our Sprime scan, we identified an introgressed region containing the haplotype defined by Zeberg and Pääbo, 2020 as both introgressed and associated with increased risk of COVID-19-severity. The overall introgressed region as detected in our Estonian population spans ~811 kb from chr3:45,843,242–46,654,616, encompasses 16 genes (Figure 2A), and ranks in the top 2% (ranked 21/1677) of Sprime detected segments based on likelihood of introgression and in the top 5% (58/1677) of Sprime segments based on length. Its extreme length provides additional support for the fact that it is introgressed and not likely a product of incomplete lineage sorting, which is detected as seemingly introgressed tracts of significantly shorter length (Huerta-Sánchez et al., 2014). Figure 2 with 2 supplements see all Download asset Open asset Computational intersections between MPRA emVars and functional genomics datasets across the severe COVID-19 risk locus. (a) Gene locations across the locus along with boundaries of the four LD blocks (A–D), borders extended to encompass all SNPs in LD (r2 > 0.3) tested with MPRA. (b) Severe COVID-19 GWAS effect sizes from release 5 of the COVID-19 Host Initiative dataset (2021), with strongest genome-wide p-values in yellow spanning the A and B blocks. See key for other color definitions. Dots and diamonds across the panels indicate respectively SNPs identified directly by Sprime (dots) and SNPs in linkage disequilibrium (r2 >0.3) with them (diamonds). (c) eQTL effect sizes across the locus (blue for CCR1, green for CCR5) in whole blood from eQTLGen (Võsa et al., 2018) across the locus. Note the strong down- versus up-regulation of CCR1 for variants in the A versus B blocks, respectively. Grey SNPs are not eQTLs for any of the two genes or were absent from the eQTL study. Asterisks denote trans-eQTLs. (d) Chromatin-based functional annotations across the locus consisting of Hi-C contacts with CCR1 and CCR5 in Spleen, Thymus, or LCL (Jung et al., 2019) and candidate cis-regulatory elements from Moore et al., 2020. (e) -Log p-values for emVars identified using MPRA across the locus. Grey SNPs failed the test for activity in either the archaic or non-archaic form. Vertical lines connect the four putative causal emVars and the most cited tag SNP rs10490770 to functional genomics and genetics data. The four putatively causal variants are unique in having significant hits across all functional genomics and genetics tests. To examine how the introgressed segment may be affecting COVID-19 severity, we began by examining the LD structure within the segment and identified four major blocks defined as minimum pairwise LD between Sprime-identified variants within a block (min r2=0.34) (Figure 2—figure supplement 1). (Please note, Figure 2 includes SNPs linked to Sprime variants whereas here we are exclusively discussing SNPs directly identified in the Sprime scan). We labeled these blocks as ‘A’ from rs13071258 to rs13068572 (chr3:45,843,242–46177096), ‘B’ from rs17282391 to rs149588566 (chr3:46,179,481–46,289,403), ‘C’ rs71327065 to rs79556692 (chr3:46,483,630–46,585,769), and ‘D’ from rs73069984 to rs73075571 (chr3:46593568–46649711) (Figure 2A; Figure 2—figure supplement 1). All Sprime alleles in the A block are significantly (p<5*10–8) associated with increased risk for COVID-19 severity (COVID-19 Host Genetics Initiative, 2021), with the median p-value being 2.32 * 10–26 and median effect size being 0.42 (Figure 2B). The B block also harbors many alleles (81.2%) significantly associated with COVID-19 severity, with the median p-value being 1.94*10–9 and median effect size being 0.28 (Figure 2B). In the C and D block, no alleles are significantly associated with COVID-19 severity, suggesting that the most likely causal variants for the COVID-19 severity association are found within the A or B blocks (Figure 2B). All the 361 Sprime-identified introgressed variants act as eQTLs in the whole blood (Võsa et al., 2018) including for many genes that are relevant to COVID-19 infection. Strikingly, of the four trans-eQTLs identified genome-wide in our Sprime scan regions in Estonians, two were located on the introgressed COVID-19 severity haplotype. These two variants, rs13063635 and rs13098911, have 11 and 33 response genes, respectively (Supplementary file 1g). We examined whether these response genes have any relevance to COVID-19 infection and found that 3 (27%) and 13 (39%) of the response genes for rs13063635 and rs13098911, respectively, are differentially expressed in at least one experiment in which a lung related cell-line or tissue was infected with COVID-19 or other related infections (Supplementary file 1h and Supplementary file 1i). These results suggest that these two trans-eQTLs may affect the lung response to COVID-19 in a way that could contribute to differential severity in host response. Furthermore, all 361 variants, including the two trans-eQTLs, act as cis-eQTLs in whole blood, altering the expression of 14 response genes: CCR1, CCR2, CCR3, CCR5, CCR9, CCRL2, CXCR6, FLT1P1, LRRC2, LZTFL1, RP11-24F11.2, SACM1L, SCAP, and TMIE. Of these genes, 7 are chemokine receptor genes (CCR1, CCR2, CCR3, CCR5, CCR9, CCRL2, and CXCR6), which are likely linked to the segment’s association with COVID-19 severity. Recent work has focused on pinpointing which of the aforementioned genes mediate(s) this risk signal, identifying CXCR6 (Schmiedel et al., 2021; Pairo-Castineira et al., 2021; Kasela et al., 2021, COVID-19 Host Genetics Initiative, 2021), CCR9 (Schmiedel et al., 2021; Kousathanas et al., 2022), SLC6A20 (Kasela et al., 2021, COVID-19 Host Genetics Initiative, 2021; Kousathanas et al., 2022), FYCO1 (Schmiedel et al., 2021), LZFL1 (Kousathanas et al., 2022), and CCR2 and CCR3 (Pairo-Castineira et al., 2021) through bayesian fine-mapping, colocalization analyses, and transcriptome-wide association (TWAS). Although these studies focused on genes physically closer to the lead risk variant (rs10490770), epigenomic dissection and functional mapping also implicated CCR1,2,3,5 genes (Stikker et al., 2022) which are farther but still deeply embedded in the introgressed haplotype. The association between COVID-19 phenotypes and CCR1 and CCR5 in particular also finds support from expression studies where elevated CCR1 expression in neutrophils and macrophages has been detected in patients with critical COVID-19 illness (Chua et al., 2020), in biopsied lung tissues from COVID-19 infected patients (Supplementary file 1h and Supplementary file 1i), as well as in Calu3 cells directly infected with COVID-19 (Supplementary file 1h and Supplementary file 1i). Likewise, elevated CCR5 expression has been detected in macrophages of patients with critical COVID-19 illness (Chua et al., 2020). Notably, some ligands for CCR1 and CCR5 (CCL15, CCL2, and CCL3) also show over-expression in these patients (Chua et al., 2020). CCRL2, LZTFL1, SCAP, and SACM1L are also differentially expressed in at least one experiment that measures differential expression of genes in lung tissues and related cell lines infected with COVID-19 or other viruses that stimulate similar immune responses (Supplementary file 1h and Supplementary file 1i). Intriguingly, when considering the effect of the cis-eQTLs for CCR1 across the entire segment, we find that the majority of alleles along the introgressed haplotype within the A block are associated with its down-regulation (average Z score = –12.3) (Figure 2C). On the other hand, the majority of alleles within the B and C blocks are associated with CCR1 up-regulation (average Z scores = 7.1 and 10.2, respectively) (Figure 2C). It is important to note that these eQTL effects are determined based on whole blood from non-infected, healthy patients (Võsa et al., 2018). When considered in the context that severe COVID-19 phenotype is characterized by increased expression of CCR1 (Chua et al., 2020), these risk-associated alleles having different directions of effect suggest that a complex change to the CCR1 regulatory landscape driven by alleles across the introgressed segment may be contributing to the disease phenotype. When we consider CCR5 expression, it shows a more consistent pattern in which the majority of alleles within the A-C blocks are associated with its down-regulation. This result is interesting as CCR5 expression in patients with severe COVID-19 illness is higher than those with more moderate cases (Chua et al., 2020). However, given the strong LD within each of these segments, discerning the direct connection between one or more alleles driving these regulatory changes and the molecular and phenotypic signatures of severe COVID-19 remains difficult. MPRA variant selection and study design To independently assess the regulatory impact of the alleles on this COVID-19 risk haplotype, we employed a Massively Parallel Reporter Assay (MPRA) to investigate which alleles on the introgressed haplotype directly affect gene expression. Alleles which have the ability to modulate gene expression in this reporter assay are candidate putatively functional alleles that may drive the association with COVID-19 severity by altering the expression of genes that facilitate the biological response to COVID-19. To ensure that we tested any potential risk variants on the haplotype, we included in the MPRA all variants directly identified in the Sprime scan as being within the introgressed COVID-19 severity associated segment (361), along with any allele linked (r2 >0.3) to one of these Sprime alleles in the Estonian population (140 alleles) or any 1000 Genomes (Auton et al., 2015) European (150 alleles) or South Asian population (197 alleles). Therefore, here we are testing not only alleles on the introgressed haplotype that have a confirmed Neanderthal-specific origin, but also alleles along the introgressed haplotype that were either already present in the human population when the haplotype was introgressed, or arose anew in humans (i.e., human-derived alleles) on the introgressed haplotype following its introgression. After filtering for SNPs falling within simple repeat regions (Benson, 1999), which are not compatible with MPRA (Tewhey et al., 2016), we identified a total of 613 experimental variants. Of these variants, 293 are significantly (p<5*10–8) associated with COVID-19 severity and another 15 approach significance (p<5*10–6), whereas 118 were not tested in the original GWAS (The COVID-19 Host Genetics Initiative Release 5) (Figure 2B). We conducted this assay in K562 cells, a leukemia cell line that displays multipotent hematopoietic biology, which allows for comparison between the MPRA data and the eQTLs identified on whole blood samples (see above). Furthermore, K562 cells can be induced into immune cell fates highly relevant to the COVID-19 severity phenotypes including monocyte, macrophage, and neutrophils (Tabilio et al., 1983; Sutherland et al., 1986; Butler and Hirano, 2014). Moreover, as K562 cells robustly grow and are transfectable using MPRA reagents, they permit the rapid, repeated acquisition of large numbers of cells, as observed in prior MPRAs (Ulirsch et al., 2016; Ernst et al., 2016). Finally, the availability of other published datasets generated on K562 cells (e.g., chromatin ChIP-seq data), allows for comparison between MPRA results, which are episomal in nature, and the endogenous behavior of the genome in the same cell type. However, we do note that MPRA results will also be limited as they will not directly reflect the response of alleles in the endogenous genome and within the in vivo tissues in which the COVID-19 response occurs. We therefore also integrated the MPRA results with datasets derived on endogenous immune tissues/cells to help improve our ability to identify biologically relevant candidate driver variants. MPRA reveals 20 expression modulating variants (emVars) We built the MPRA library following Tehwey and colleagues (2016) and performed four replicates of the experiment in K562 cells (Methods). We observed that normalized transcript counts between replicates were highly correlated (Pearson’s R>0.999 p=p-value<2.2e-16) (Figure 3A). As with other MPRA studies (Tewhey et al., 2016; Uebbing et al., 2021), transcript counts in the cDNA samples are significantly correlated with, but not completely explained by sequence representation in the DNA plasmid pool (correlation between means: Pearson’s R=0.24 p=2.2e-16; Spearman’s ρ=0.912 p-value <2.2e-16) (Figure 3B), suggesting that while some sequences do not have an effect on transcription, other do. The expression of positive control sequences in this experiment was significantly correlated with their expression in the source MPRA (Pearson’s R=0.59, p=0.0058) (Figure 3C). Any deviation between positive control sequence activity in this assay and the source MPRA for the two control sets is likely due to additional regulatory information in this assay for which the tested sequences are 270 bp compared with 170 bp in the source MPRA. Moreover, for our positive control set we observed that 95% of control sequences displayed activity, whereas only 14% of the negative control sequences displayed activity (Figure 3D). Figure 3 Download asset Open asset MPRA results show reproducibility and accuracy. (a) Log normalized counts for each tested sequence in replicate 1 compared with the replicate 2 of the MPRA. Pearson’s R and Spearman’s ρ are extremely high and significant across pairwise replicate comparisons of all four replicates (R>0.99 p-value <2.2 *10–16; ρ=0.98 p-value <2.2 *10–16). (b) Log normalized sequence counts for each tested in the plasmid DNA averaged across the four replicates compared with log normalized average sequence counts in the cDNA averaged across the four replicates. As with other MPRA studies (Tewhey et al., 2016; Uebbing et al., 2021), there is a significant correlation but the plasmid counts do not fully explain the cDNA counts (Pearson’s R=0.24 p-value <2.2 *10–16; Spearman’s ρ=0.92 p-value <2.2 *10–16), suggesting that some of the sequences have an effect on transcription. Sequences determined to be significantly active in the MPRA (methods) are colored in red, non-significant points are black. (c) Activity log fold change (LFC cDNA:pDNA) of positive control sequences in the source MPRA (Jagoda et al., 2021) and in this MPRA. The significant correlation (Pearson’s R=0.57 p-value = 0.006; Spearman’s ρ=0.51 p-value 0.016) suggests that the activity results in this MPRA are accurate. (d) Fraction of sequences tested showing significant activity (LFC cDNA:pDNA corrected p-value >0.01). 95% of positive control sequences tested and 0.14% of negative control sequences tested show activity once again suggesting accuracy in the MPRA results. 53% of experimental sequences show significant activity. Of the 613 experimental (1,226 alleles) tested variants, 327 (53%) were within sequences found to have detected effects on reporter gene expression (i.e., they are considered ‘active’ or cis-regulatory elements [CREs]) in the context of either the allele on the introgressed haplotype or via its alternative variant (Figure 3D) (Supplementary file 1j). Consistent with other MPRA studies (Tewhey et al., 2016; Ulirsch et al., 2016; Uebbing et al., 2021), most active sequences showed relatively small effects, with only 17.1% of active sequences showing a log fold change (LFC) greater than 2 (Figure 4A). To confirm that these active CRE sequences reflect endogenous K562 biology, we compared the distribution of active CREs with K562 chromatin state data (Ernst and Kellis, 2017; Sloan et al., 2016). We observed that active CREs are significantly enriched relative to non-active sequences for falling with K562 DNase I Hypersensitivity Sites (DHS) and within poised promoters (OR: 8.05, p=0.023) (Figure 4B). They are also borderline significantly depleted of falling within heterochromatin (OR: 0.73, p=0.072) (Figure 4B). Figure 4 Download asset Open asset Properties of MPRA-identified active sequences and expression modulating variants. (a) Log Fold Change of the cDNA count compared to the plasmid DNA for each sequence and the -log10 associated multiple hypothesis corrected p-value. Active sequences are those with a corrected p-value <0.01; this threshold is denoted with a blue dashed line. The larger plot has a y-axis limit of 20; the inset on the right shows the full spread of the data with the light red shaded box denoting the area shown in the larger figure to the left. (b,d) Enrichment of active sequences in K562 genomic features relative to non-active sequences (b) or sequences with emVars relative to sequences without emVars (d). Genomic features indicated with a number represent chromatin states in K562 cells as defined by Ernst and Kellis, 2017. DHS and H3K27ac derive from ENCODE (Butler and Hirano, 2014). Enrichments are reported as Fisher’s odds ratios with lines indicating confidence intervals. Significant enrichments (p<0.05) are colored in red. Missing chromatin states had no overlap with either active sequences (b) or those containing emVars (d). (c) Log Fold Change between active sequences with the allele on the introgressed haplotype compared with the sequence containing the other allele. Expression modulating variants (emVars) are those whose LFC for this measure is significant with a corrected p-value <0.01; this threshold is denoted with a blue dashed line. We next defined as ‘expression modulating variants’ (emVars) those variants exhibiting a significant difference of expression between their two allelic versions using a multiple hypothesis corrected p-value less than 0.01. Using this approach, we identified 20 emVars among the 613 variants we tested (Table 1, Figure 2E, Figure 3B). Consistent with previous MPRA studies (Tewhey et al., 2016; Ulirsch et al., 2016), the effect sizes of most emVars detected here are relatively modest, with only 1 emVar having an absolute LFC greater than 2 (Table 1, Figure 4C). Because the sample size is quite small (20), CREs containing emVars do not show significant enrichment within any endogenous K562 functional annotations. However, compared with tested sequences that do not contain emVars, CREs containing these emVars trend toward being over-represented within poised promoters, weak enhancers, DHS, insulators, repressive marks, and for depletion in heterochromatin regions (See Figure 4D). Table 1 Properties of emVars and prioritization. This table shows a summary of all the functional data we obtained on the 20 significant emVars identified by the MPRA which we used for prioritization of which these 20 emVars show the most evidence for contributing to the severe COVID-19 phenotype. Specifically, we looked for (1) concordance between the eQTL data (Columns H,I) and the Hi-C Data (K) - specifically for emVars that are eQTLs for a COVID-19 relevant gene with which they also physically interact in an immune tissue; (2) a significant association between the allele and severe COVID-19 in the GWAS data (column J); (3) overlap between the emVar and an ENCODE annotated cCRE (column L) with support in at least one ‘class A’ immune tissue (column M). The 4 variants that met all these criteria are highlighted in bold. For visualiza" @default.
• W4320078097 created "2023-02-12" @default.
• W4320078097 creator A5077936005 @default.
• W4320078097 date "2021-07-16" @default.
• W4320078097 modified "2023-10-02" @default.
• W4320078097 title "Editor's evaluation: Regulatory dissection of the severe COVID-19 risk locus introgressed by Neanderthals" @default.
• W4320078097 doi "https://doi.org/10.7554/elife.71235.sa0" @default.
• W4320078097 hasPublicationYear "2021" @default.
• W4320078097 type Work @default.
• W4320078097 citedByCount "0" @default.
• W4320078097 crossrefType "peer-review" @default.
• W4320078097 hasAuthorship W4320078097A5077936005 @default.
• W4320078097 hasBestOaLocation W43200780971 @default.
• W4320078097 hasConcept C104317684 @default.
• W4320078097 hasConcept C142724271 @default.
• W4320078097 hasConcept C2779134260 @default.
• W4320078097 hasConcept C3008058167 @default.
• W4320078097 hasConcept C524204448 @default.
• W4320078097 hasConcept C54355233 @default.
• W4320078097 hasConcept C71924100 @default.
• W4320078097 hasConcept C84597430 @default.
• W4320078097 hasConcept C86803240 @default.
• W4320078097 hasConceptScore W4320078097C104317684 @default.
• W4320078097 hasConceptScore W4320078097C142724271 @default.
• W4320078097 hasConceptScore W4320078097C2779134260 @default.
• W4320078097 hasConceptScore W4320078097C3008058167 @default.
• W4320078097 hasConceptScore W4320078097C524204448 @default.
• W4320078097 hasConceptScore W4320078097C54355233 @default.
• W4320078097 hasConceptScore W4320078097C71924100 @default.
• W4320078097 hasConceptScore W4320078097C84597430 @default.
• W4320078097 hasConceptScore W4320078097C86803240 @default.
• W4320078097 hasLocation W43200780971 @default.
• W4320078097 hasOpenAccess W4320078097 @default.
• W4320078097 hasPrimaryLocation W43200780971 @default.
• W4320078097 hasRelatedWork W1828691184 @default.
• W4320078097 hasRelatedWork W1982713802 @default.
• W4320078097 hasRelatedWork W1986695036 @default.
• W4320078097 hasRelatedWork W1991523530 @default.
• W4320078097 hasRelatedWork W2000479887 @default.
• W4320078097 hasRelatedWork W2002128513 @default.
• W4320078097 hasRelatedWork W2057739827 @default.
• W4320078097 hasRelatedWork W3146600830 @default.
• W4320078097 hasRelatedWork W3162081874 @default.
• W4320078097 hasRelatedWork W2092874662 @default.
• W4320078097 isParatext "false" @default.
• W4320078097 isRetracted "false" @default.
• W4320078097 workType "peer-review" @default.