Post by Admin on Jun 4, 2024 1:46:28 GMT
Collection of functional variants
A particular interest in this study is the distribution of functional variation, not in the least due to the potential impact on phenotypes, especially to those with medical relevance [27]. As much as 97.5% of all annotated variation was discovered outside of the known functional elements (upstream, downstream, intron, and intergenic). These results are similar to the expected distributions of mutations shown with the simulated data [28]. Nevertheless, there were >8,000 mutations discovered within exons of each individual on average (top half of Table 2). We annotated several classes of functional mutations within the coding regions (bottom half of Table 2). As expected, the nonsense mutations classified in the annotation file as “disruptive in-frame indel,” “start lost,” “stop gained,” and “stop loss” were rare, while categories with minimal effect on the function, such as “synonymous,” “motiff,” “protein folding,” and “missense,” were more common. Some of the mutations listed in the annotation filecan be classified in >1 category (e.g., “synonymous variants” can also be counted in “exonic variants”).
In addition to the gene-coding mutations, we report a number of regulatory variants. For example, the database contains a total of 2,229 transcription factor binding site ablation (TFBS) mutations (bottom half of Table 2). A summary of functional variation discovered in this study is presented in Table 2. The full list of high-impact functional variants (including frameshift, start lost/stop lost or gained, transcript ablations, and splice alterations) that had an allele count of 2 or more with their predicted function, number of gene transcripts of the gene affected, and frequencies is presented in Supplementary Table S3. The full annotation database with classifications is available alongside the associated data deposited in GigaDB [11].
Collection of the medically relevant variants
Many of the reported variants are already known to be medically related and are listed either in genome-wide association studies (GWAS) [29] or ClinVar (an NCBI archive of reports of the relationships among human variations and phenotypes with supporting evidence) [30] catalogues (Table 3). Our database contains a total of 43,892 benign mutations in medically related genes but also 189 unique pathogenic or likely pathogenic variants, as well as 20 protective or likely protective alleles as defined in ClinVar [30, 31]. Each individual in this study carries 19 pathogenic and 12 protective mutations on average. While at least some individuals were homozygous for the pathogenic allele, none of the associated disease phenotypes have been reported, which could be largely attributed to heterozygosity, age-dependent penetrance, expressivity, and gene-by-environment interactions [32, 33].
Table 3:Medically relevant variants in the Ukrainian population included in GWAS [29] and ClinVar [30] databases
Source of annotation No. unique substitutionsa Total allele No. Mean per sample
GWAS catalog 102,551 6,479,953 66,804
ClinVar: pathogenic (or likely pathogenic) 189 1,830 19
ClinVar: benign (or likely benign) 43,892 1,842,668 18,997
ClinVar: protective (or likely protective) 20 1,209 12
a
Unique variants represent substitutions that were counted only once, disregarding their frequency in the population.
As expected, our study shared a lot more variants with the GWAS [29] than with the ClinVar [30] catalogue. While GWAS has recently become the tool of choice to identify genetic variants associated with complex disease and other phenotypes of interest [34], because the amount of genetic variance explained by these variants is low, they are generally not very useful for predicting pathogenic phenotypes [35]. It is also important to note that not all ClinVar variants carry the same weight of supporting evidence; attributing disease causation to prioritized variants remains an inexact process and some of the reported associations eventually are proven to be spurious [36]. Nevertheless, the importance of the unique set of mutations published here is difficult to overemphasize because it constitutes the first published set of pathological variants in an understudied population, an important step towards a local catalogue of medically relevant mutations. In addition, as the attention in the genomic community is shifting from monogenic to polygenic traits, many of these may become relevant in future research and exploration [37]. A full list of the medically relevant functional markers found in the Ukrainian population and reported in GWAS [29] and ClinVar [30] databases is presented in Supplementary Table S4 with alternative allele frequencies and annotations.
Disease variants with frequencies that differed between the Ukrainians and the neighboring populations are of particular interest to the medical community. It is well established that differences in allele frequencies are a consequence of evolutionary forces acting in populations (such as drift, mutation, migration, nonrandom mating, and natural selection) and that certain diseases and heritable traits display marked differences in frequency between populations [38]. With this in mind, we created a list of the known disease variants whose frequencies differ between Ukrainians and other European populations (the combined European sample [EUR] from the 1KG, comprising Utah residents [CEU] with Northern and Western European ancestry, Toscani in Italy [TSI], Finnish in Finland [FIN], British in England and Scotland [GBR], Iberian population in Spain [IBS] [14, 39], and French population from Human Genome Diversity Project [HGDP] [FRA] [40]) and Russians from HGDP (RUS) [40]. Several examples of these variants are presented in Table 4. Among these are variants involved in a number of medical conditions such as hyperglycinuria/iminoglycinuria (rs35329108,SLC6A19), efficacy of bisphosphonate response (rs2297480,FDPS), autism (rs7794745, CNTNAP2), Leber congenital amaurosis (rs10151259,RPGRIP1), and breast cancer susceptibility in BRCA1 and BRCA2 carriers (rs1801320,RAD51) (Table 4).
Table 4:Examples of the functional SNPs with highly differentiating functional markers reported in ClinVar [29] with high differences in the Ukrainian population compared to other neighboring European populations
SNP Chr Gene REF/alta Associated medical condition Non-reference allele frequency Fisher exact test P-value
UKR EUR RUS vs EUR vs RUS
rs2297480 1 FDPS T/G Efficacy of the bisphosphonate response 0.13 0.26 0.27 0.038 >0.001
rs35329108 5 SLC6A19 G/A Hyperglycinuria, iminoglycinuria 0.32 0.23 0.17 0.049 0.004
rs7794745 7 CNTNAP2 A/T Autism 0.48 0.38 0.30 0.032 0.010
rs10151259 14 RPGRIP1 G/T Leber congenital amaurosis, cone-rod dystrophy 0.32 0.24 0.11 0.003 0.014
rs1801320 15 RAD51 G/C Breast cancer susceptibility inBRCA1andBRCA2carriers 0.19 0.08 0.07 0.047 0
a
The reference allele (REF) is set according to the reference allele in GrCH38.p13 [18]. Other European populations consist of the combined sample from Western and Central Europe from 1KG (EUR) [14, 39] and French population from HGDP (FRA) [40], as well as Russians (RUS) from HGDP [40]. Non-reference allele frequency is reported compared to the reference allele in GRCh38. Differences are evaluated by the Fisher exact test. All the functional SNPs with significant population frequency differences are listed in Supplementary Table S5.
Of course, not all the medically related variants are currently known, and many remain to be discovered and verified in local populations. This is, to some extent, a consequence of underreporting of allelic endemism within understudied populations, particularly in Eastern Europe [10] but also elsewhere [41, 42]. By offering public annotations of functional mutations in a population sampled across the territory of Ukraine, our database contributes a number of candidates to direct future research in medical genomics. We chose only the markers with the highest non-reference allele frequency differences compared to the neighboring populations EUR [14] and RUS [40], evaluated by the Fisher exact test, and listed them in Table 5.
Table 5:Examples of the functional markers with the highest non-reference allele frequency differences in the Ukrainian population evaluated by the Fisher exact test compared to the frequencies in the neighboring populations: the combined population from Europe (EUR) [14] and Russians from HGDP (RUS) [40]
SNP Chr Gene Ref/Alt Function Non-reference allele frequency Fisher exact test P-value
UKR EUR RUS vs CEU vs RUS
rs72625995 17 POM121L8P C/T Exonic, nonsynonymous SNV 0.03 0.62 0.75 2.50E−07 1.86E−06
rs9930886 16 PTPRN2 A/G Exonic, synonymous SNV 0.01 0.33 0.35 2.56E−07 2.19E−06
rs4779816 15 ZBTB9; BAK1 A/G Exonic, nonsynonymous SNV 0.41 0.80 0.83 3.29E−06 7.82E−07
rs58580222 12 ABCC1 G/A Exonic, synonymous SNV 0.03 0.13 0.26 3.06E−04 1.17E−02
rs80150964 11 SMIM40; KIFC1 T/C Exonic, non-synonymous SNV 0.03 0.23 0.19 4.95E−04 1.96E−06
A particular interest in this study is the distribution of functional variation, not in the least due to the potential impact on phenotypes, especially to those with medical relevance [27]. As much as 97.5% of all annotated variation was discovered outside of the known functional elements (upstream, downstream, intron, and intergenic). These results are similar to the expected distributions of mutations shown with the simulated data [28]. Nevertheless, there were >8,000 mutations discovered within exons of each individual on average (top half of Table 2). We annotated several classes of functional mutations within the coding regions (bottom half of Table 2). As expected, the nonsense mutations classified in the annotation file as “disruptive in-frame indel,” “start lost,” “stop gained,” and “stop loss” were rare, while categories with minimal effect on the function, such as “synonymous,” “motiff,” “protein folding,” and “missense,” were more common. Some of the mutations listed in the annotation filecan be classified in >1 category (e.g., “synonymous variants” can also be counted in “exonic variants”).
In addition to the gene-coding mutations, we report a number of regulatory variants. For example, the database contains a total of 2,229 transcription factor binding site ablation (TFBS) mutations (bottom half of Table 2). A summary of functional variation discovered in this study is presented in Table 2. The full list of high-impact functional variants (including frameshift, start lost/stop lost or gained, transcript ablations, and splice alterations) that had an allele count of 2 or more with their predicted function, number of gene transcripts of the gene affected, and frequencies is presented in Supplementary Table S3. The full annotation database with classifications is available alongside the associated data deposited in GigaDB [11].
Collection of the medically relevant variants
Many of the reported variants are already known to be medically related and are listed either in genome-wide association studies (GWAS) [29] or ClinVar (an NCBI archive of reports of the relationships among human variations and phenotypes with supporting evidence) [30] catalogues (Table 3). Our database contains a total of 43,892 benign mutations in medically related genes but also 189 unique pathogenic or likely pathogenic variants, as well as 20 protective or likely protective alleles as defined in ClinVar [30, 31]. Each individual in this study carries 19 pathogenic and 12 protective mutations on average. While at least some individuals were homozygous for the pathogenic allele, none of the associated disease phenotypes have been reported, which could be largely attributed to heterozygosity, age-dependent penetrance, expressivity, and gene-by-environment interactions [32, 33].
Table 3:Medically relevant variants in the Ukrainian population included in GWAS [29] and ClinVar [30] databases
Source of annotation No. unique substitutionsa Total allele No. Mean per sample
GWAS catalog 102,551 6,479,953 66,804
ClinVar: pathogenic (or likely pathogenic) 189 1,830 19
ClinVar: benign (or likely benign) 43,892 1,842,668 18,997
ClinVar: protective (or likely protective) 20 1,209 12
a
Unique variants represent substitutions that were counted only once, disregarding their frequency in the population.
As expected, our study shared a lot more variants with the GWAS [29] than with the ClinVar [30] catalogue. While GWAS has recently become the tool of choice to identify genetic variants associated with complex disease and other phenotypes of interest [34], because the amount of genetic variance explained by these variants is low, they are generally not very useful for predicting pathogenic phenotypes [35]. It is also important to note that not all ClinVar variants carry the same weight of supporting evidence; attributing disease causation to prioritized variants remains an inexact process and some of the reported associations eventually are proven to be spurious [36]. Nevertheless, the importance of the unique set of mutations published here is difficult to overemphasize because it constitutes the first published set of pathological variants in an understudied population, an important step towards a local catalogue of medically relevant mutations. In addition, as the attention in the genomic community is shifting from monogenic to polygenic traits, many of these may become relevant in future research and exploration [37]. A full list of the medically relevant functional markers found in the Ukrainian population and reported in GWAS [29] and ClinVar [30] databases is presented in Supplementary Table S4 with alternative allele frequencies and annotations.
Disease variants with frequencies that differed between the Ukrainians and the neighboring populations are of particular interest to the medical community. It is well established that differences in allele frequencies are a consequence of evolutionary forces acting in populations (such as drift, mutation, migration, nonrandom mating, and natural selection) and that certain diseases and heritable traits display marked differences in frequency between populations [38]. With this in mind, we created a list of the known disease variants whose frequencies differ between Ukrainians and other European populations (the combined European sample [EUR] from the 1KG, comprising Utah residents [CEU] with Northern and Western European ancestry, Toscani in Italy [TSI], Finnish in Finland [FIN], British in England and Scotland [GBR], Iberian population in Spain [IBS] [14, 39], and French population from Human Genome Diversity Project [HGDP] [FRA] [40]) and Russians from HGDP (RUS) [40]. Several examples of these variants are presented in Table 4. Among these are variants involved in a number of medical conditions such as hyperglycinuria/iminoglycinuria (rs35329108,SLC6A19), efficacy of bisphosphonate response (rs2297480,FDPS), autism (rs7794745, CNTNAP2), Leber congenital amaurosis (rs10151259,RPGRIP1), and breast cancer susceptibility in BRCA1 and BRCA2 carriers (rs1801320,RAD51) (Table 4).
Table 4:Examples of the functional SNPs with highly differentiating functional markers reported in ClinVar [29] with high differences in the Ukrainian population compared to other neighboring European populations
SNP Chr Gene REF/alta Associated medical condition Non-reference allele frequency Fisher exact test P-value
UKR EUR RUS vs EUR vs RUS
rs2297480 1 FDPS T/G Efficacy of the bisphosphonate response 0.13 0.26 0.27 0.038 >0.001
rs35329108 5 SLC6A19 G/A Hyperglycinuria, iminoglycinuria 0.32 0.23 0.17 0.049 0.004
rs7794745 7 CNTNAP2 A/T Autism 0.48 0.38 0.30 0.032 0.010
rs10151259 14 RPGRIP1 G/T Leber congenital amaurosis, cone-rod dystrophy 0.32 0.24 0.11 0.003 0.014
rs1801320 15 RAD51 G/C Breast cancer susceptibility inBRCA1andBRCA2carriers 0.19 0.08 0.07 0.047 0
a
The reference allele (REF) is set according to the reference allele in GrCH38.p13 [18]. Other European populations consist of the combined sample from Western and Central Europe from 1KG (EUR) [14, 39] and French population from HGDP (FRA) [40], as well as Russians (RUS) from HGDP [40]. Non-reference allele frequency is reported compared to the reference allele in GRCh38. Differences are evaluated by the Fisher exact test. All the functional SNPs with significant population frequency differences are listed in Supplementary Table S5.
Of course, not all the medically related variants are currently known, and many remain to be discovered and verified in local populations. This is, to some extent, a consequence of underreporting of allelic endemism within understudied populations, particularly in Eastern Europe [10] but also elsewhere [41, 42]. By offering public annotations of functional mutations in a population sampled across the territory of Ukraine, our database contributes a number of candidates to direct future research in medical genomics. We chose only the markers with the highest non-reference allele frequency differences compared to the neighboring populations EUR [14] and RUS [40], evaluated by the Fisher exact test, and listed them in Table 5.
Table 5:Examples of the functional markers with the highest non-reference allele frequency differences in the Ukrainian population evaluated by the Fisher exact test compared to the frequencies in the neighboring populations: the combined population from Europe (EUR) [14] and Russians from HGDP (RUS) [40]
SNP Chr Gene Ref/Alt Function Non-reference allele frequency Fisher exact test P-value
UKR EUR RUS vs CEU vs RUS
rs72625995 17 POM121L8P C/T Exonic, nonsynonymous SNV 0.03 0.62 0.75 2.50E−07 1.86E−06
rs9930886 16 PTPRN2 A/G Exonic, synonymous SNV 0.01 0.33 0.35 2.56E−07 2.19E−06
rs4779816 15 ZBTB9; BAK1 A/G Exonic, nonsynonymous SNV 0.41 0.80 0.83 3.29E−06 7.82E−07
rs58580222 12 ABCC1 G/A Exonic, synonymous SNV 0.03 0.13 0.26 3.06E−04 1.17E−02
rs80150964 11 SMIM40; KIFC1 T/C Exonic, non-synonymous SNV 0.03 0.23 0.19 4.95E−04 1.96E−06