20090119 SNP

Result

SNP identification

A total of 2,799564 SNPs are found from the KSJ genome as defined by MAQ x software and additional filteration, wich criteria were described in Supplementary Information (Figure X). Evaluation of the quantity and quality of the identified variants was undergone by comparing the overlap and uniqueness of these variants with know SNPs in dbSNP ver. 129(Figure X).

We assessed our SNP calling accuracy by comparing the identified SNPs in the YH sequence with dbSNP. We detected that 2,477,752 of KSJ variants were presented in dbSNP (2,158,552 as validated and 319,200 as non-validated SNPs) and the remaiing 321,812 SNPs were revealed as the novel (Figure X). Among validated SNPs, 40000 were located in genic regions: xxx were in 5' UTR, yyy in 3' UTR, and zzz in CDS regions, in which ttt were non-synonymous SNPs. In case of non-validated variants, xxx were located in 5' UTR, yyy in 3' UTR, and zzz in CDS region, and ttt were identified as non-synonymous from CDS SNPs. Of the detected 321,812 novel variants, only about 1.4% of these were presented whithin the known gene regions: 2,381 variants were in 5' UTR, 10,920 in 3' UTR, and 17,585 in CDS with 9,472 non-synonymous SNPs (Figure X).

In addition to the comparing with dbSNP, the KSJ SNPs were comapred with the three available individual genomes of James D. Watson, J. Craig Venter, and BGI YG x-y. Among four genomes,

e three individuals also have a similar fraction of non-synonymous SNPs (YH, 7,062 (0.23%); Venter, 6,889 (0.22%); Watson, 7,319 (0.20%)). There are 2,622 non-synonymous SNPs shared among the three individuals, accounting for 37.1% of non-synonymous SNPs in the YH genome.

. In looking at the SNPs of the three individual genomes, all share 1.2 million SNPs. Each also has a set of SNPs unique to their own genome: for YH, 978,370 (31.8%) SNPs; for Venter, 924,333 (30.1%); and for Watson, 1,096,873 (33.0%) (Supplementary Fig. 2).

The three individuals also have a similar fraction of non-synonymous SNPs (YH, 7,062 (0.23%); Venter, 6,889 (0.22%); Watson, 7,319 (0.20%)). There are 2,622 non-synonymous SNPs shared among the three individuals, accounting for 37.1% of non-synonymous SNPs in the YH genome.

Validation of SNPs by experimental genotyping

The accuray of SNP calling was assessed by comparing the identifed SNPs in the KSJ genome sequence with genotype data from Affymetrix 550K and Illumina 1M duo chips (Figure X). 각 genotype chip으로부터 450K 및 950K의 SNP calls을 obstained했고, KSJ SNPs와는 각각 300K (78%) and 700K (76%)가 concordance를 가졌다. 이중에서, 100,000개 정도가 Three SNP data들에서 공통적으로 share 되었다. Three SNP 데이터중, 약 30%가 discordant 했다. 원인은 ?

We used polymerase chain reaction (PCR) amplification and traditional Sanger sequencing technology on a subset of the inconsistent SNPs and small indels to determine whether they conformed to the genotyping or GA sequencing results (Supplementary Table 2). Of the 50 SNPs examined, 82.0% (41 SNPs) were consistent with the GA sequencing, indicating that the YH genome has a 99.98% accuracy over these genotyped sites (Supplementary Table 3). We also validated 100% of the PCR-amplified YH genome non-coding-region indels and 90% of the frameshift indels (Supplementary Table 4).

Known phenotypic or disease risk variant screen

The primary goal of personal genome sequencing is to allow identification of disease risk genotypes. We surveyed 1,495 alleles of 116 genes in the YH genome in the Online Mendelian Inheritance in Man (OMIM)¹⁸ database and found one mutation in the GJB2 gene, which is associated with a recessive deafness disorder. This allele was heterozygous, thus there was no expectation of, or evidence for, deafness in this individual, but it does raise the possibility of offspring having this disorder.

A preliminary search of genes and variants associated with common, complex phenotypes or disorders using OMIM data (Table 3) identified several genotypes that confer risk for tobacco addiction and Alzheimer's disease. This donor is a heavy smoker, as is consistent with individuals of similar genotypes in tobacco addiction studies. The donor contains 9 (56.3%) of the 16 identified Alzheimer's disease risk alleles³, including two APOE alleles¹⁹ and 7 SORL1 alleles²⁰. These findings indicate an increased risk for Alzheimer's disease, but there are no available data from any family members to assess whether there is a family history of Alzheimer's disease.

Methods

Public data used

The human reference genome, together with genes and repeats annotation, was downloaded from UCSC database (http://genome.ucsc.edu/), which has the same sequence as the NCBI build 36.1. The NCBI reference genes with prefix “NM” were mapped to the reference genome using BLAT by UCSC. Hits with >90% identity were retained for further analysis, and only one transcript was retained for each gene. dbSNP v128 and HapMap release 23 were used. The SNP set from Venter’s genome was downloaded from the public FTP site of JCVI (ftp://ftp.jcvi.org/pub/data/huref/), and the SNP set of Watson’s genome was provided by Baylor College of Medicine (BCM).

Annotation of SNPs and indels

SNPs and indels were compared with NCBI dbSNP v128 to distinguish known SNPs (those that had been deposited to dbSNP) and novel SNPs (those that were not in dbSNP). All the SNPs were annotated by comparing their position to other genomic features including gene regions, repeat elements, etc.

We carried out multi-alignments of cDNAs from different species (rhesus, mouse, dog, opossum, chicken, and stickleback) that covered YH indels that might cause a frameshift. If the YH sequence had the same sequence as the outgroups, we defined it as the ancestral type and considered the genes on NCBI reference to have the frameshift. In regions other than exons, we only used the chimpanzee genome as an outgroup to determine whether the YH or NCBI reference genome had the ancestral allele.

GO analysis

The gene ontology classification analysis was performed using WEGO34 (http://wego.genomics.org.cn), a tool for visualizing, comparing, and plotting GO annotation results. We used the Fisher exact test and required p values to be smaller than 0.001 for a specific category of genes in GO classifications, in order for it to be considered significantly different.

Discussion

The rate of heterozygous SNPs, which is an indication of the sequence diversity in the YH genome, is 6.94×10-4 across the autosomes (Supplementary Table 7). As estimated above, about 7% of the novel heterozygotes may have one allele missing, thus the rate calculated here is a likely to be slightly low.This difference is likely because the sequencing of YH was done using very short reads, which makes it impossible to identify long indels and unlikely to detect indels in highly repetitive regions. In the coding regions, the YH genome SNP rate (3.35×10-4) is 2.1 times lower, and the rate of small indels (0.09×10-5) is 23 times lower than the average in the whole genome.

--> 정확도 부분인데, 김태형연구원과 협의 필요

The SNP rate in 5’-UTR (4.79×10-4) and 3’-UTR (5.27×10-4) is 31% and 24% lower than in the whole genome, while the rate of small indels in 5’-UTR (0.72×10-5) is 2.3 times lower than that in 3’-UTR (2.36×10-5). By adopting the population mutation parameter, which is a measure to correct for sample size or number of chromosomes, the estimated rate of SNPs and small indels between YH and the NCBI reference is 8.06×10-4 and 3.42×10-5, respectively.

--> SNP의 genetic region 양상에 대한 토의

The frequency of heterozygous and homozygous SNPs in our dataset that are in dbSNP (validated) was nearly equal (51.8% and 48.2%), but in the set of novel SNPs, the frequency of heterozygotes was 5.7 times higher. Such a difference in frequency may be because most common alleles have already been identified and placed in dbSNP, whereas novel alleles are likely to be rare and often exist as heterozygotes. Overall, the ratio of heterozygous to homozygous SNPs in the YH autosomes is 1.34, which is lower than expected from the Hardy-Weinberg principle. This might be due to the possible existence of rare alleles in the NCBI reference genome3 and by the miss-identification of some heterozygotes in the YH genome Transitions in the YH genome SNPs are four times more frequent than transversions, but there is little obvious bias among each type of transition or transversion combination; this is the same pattern seen in a previous study on NCBI dbSNPs22

--> novel SNP중에 heterozygous SNP 수에 대한 토의, transition 과 transversion갯수 비교

7,062 non-synonymous SNPs in the YH genome and these are distributed throughout 4,293 genes. The ratio of the non-synonymous and synonymous mutation rate (dN/dS) over all these genes was 0.35. We examined the GO23 classification of genes that had a dN/dS ratio greater than 0.35, and found that these genes primarily belong to functional categories that are known to be under relaxed selection or to have a high divergence rate, these include genes that encode zinc-finger proteins, and genes related to the immune system, to antioxidant activities, and to physiological responses to stress or stimuli24. The other functional categories that also have a higher fraction of non-synonymous SNPs are genes involved in the basement membrane, proteinaceous extracellular matrix, and enzyme inhibitors (Fisher exact test p<0.001). (See complete list in Supplementary Table 8.)

--> nonsyn SNP토의 , dN/dS ratio

In addition to comparison with OMIM, we also compared YH genotypes with all available genotypes in the Human Gene Mutation Database (HGMD)33 A total of 20,559 genotypes matched with 1,478 HGMD genes, of which 318 genotypes were associated with increased disease risk (Supplementary Table 11). Many of these specific variations that are potentially associated with disease risk have not yet been tested in sufficiently large population samples or have not been surveyed in different ethnic populations to provide a good assessment of the risk inherent in the presence of these genotypes in YH. A much more extensive survey of the frequencies of the diseases associated with variants in a broad range of populations and samples will be required to validate the level of risk of disease for an individual.

--> known phenotype, disease risk

Figure S2: Overlap of (a) all SNPs, (b) non-synonymous SNPs, and (c) genes with non-synonymous SNPs among YH, Venter, and Watson genome.

Figure S7: Example of a frameshift on the NCBI reference genome where the YH genome had the same allele type as that in the sequences of all other organisms examined. The deletion on the NCBI human reference genome is present at the 580th amino acid of protein NP_001103669. That this gene can be transcribed is supported by 4 cDNAs (NCBI accession numbers: BC071856, BC096755, AK092590 and AX747617).

Table S5 Percent of small indels in the YH or the NCBI36 genome that have the same allele type as the chimpanzee genome. The identified 1–3 bp indels between YH and the NCBI reference genome were checked against the chimpanzee syntenic regions that were longer than 100-bp and greater than 95% identity. The alleles that are identical to those in the chimpanzee were taken as the ancestral allele types

Table S7: Rate of SNPs and 1–3 bp indels in complete autosomes and in defined genetic regions. The Rate between YH and the NCBI reference (“vs Ref”) was calculated using the population mutation parameter θ=K/aL, a=1+1/2+…+1/(n-1), where K is the number of variant sites found by sequencing n chromosomes in a region of length L. YH is a diploid sequence, while the NCBI genome represents one set of human reference chromosomes, so n=3.

--> table S5, S7은 SNP로 바꿔서 만들어야 함

Table S8: Full list of genes containing non-synonymous SNPs.

Table S11: List of HGMD alleles which are positive in YH genome.

------------READ SNP------------

TOTAL : 2825117

-------------------

VALID : 2441994

UTR-3: 16934

UTR-5: 3502

SPLICE-3: 24

SPLICE-5: 39

INTRON: 905210

CDS-SYNON: 9607

CDS-NONSYNON: 0

NONSENSE: 206

MISSENSE: 9394

FRAMESHIFT: 26

-------------------

UNVALID : 383123

UTR-3: 1616

UTR-5: 576

SPLICE-3: 1

SPLICE-5: 7

INTRON: 133906

CDS-SYNON: 750

CDS-NONSYNON: 0

NONSENSE: 30

MISSENSE: 848

FRAMESHIFT: 16