KR101460520B1 - Detecting method for disease markers of NGS data - Google Patents

Abstract

In order to improve biomarker search and digging power, the present invention uses a continuous variable used in defining a base polymorphism, and uses comparative analysis methods (allele ranks, allele depth tangent difference, absolute difference in alleles, The results of this study were as follows: 1) The selection and ranking of disease mutation markers based on the gene score, the statistical allele difference, the ratio of alleles imbalance, and the weight score of the locus function were made, and reverse classification was performed. The sensitivity of detecting genomic biomarkers corresponding to diseases, chronic diseases, disease risk and phenotype of normal persons is improved compared to the conventional methods.

Description

TECHNICAL FIELD [0001] The present invention relates to a method for detecting a disease marker of a next generation sequencing data,

The present invention relates to a system and method for detecting a disease variation marker in NGS data, and more particularly, to a system and method for detecting a disease variation marker in NGS data, The present invention relates to a system and method for detecting a disease variation marker of NGS data by inverse classification based on gene depth, signal intensity, and quality score.

Currently, the trend of the IT market is changing in the order of Google cloud computing Ubiquitous. At the same time, biomedical, bioinformation and genome areas are changing to new trends in order of bio-google system bio personalized medicine. In particular, the post-human genome project shows that national competition is taking place in this trend, as shown in Table 1 (list of publications related to the Post-Human Genome Project), and the recent next generation sequencing technology (NGS) Personalized medicine is becoming a reality.

Post Human Genome Project Publications One Mapping and sequencing of structural variation from eight human genomes Nature, 453, 2008 2 A plan to capture Human Diversity in 1000 Genomes Nature, 319, 2008 3 The diploid genome sequence of an Asian individual, Nature, 456 2008 4 The first Korean genome sequence and analysis: full genome sequencing for a socio-ethic group Genome Research 2009 5 A highly annotated whole-genome sequence of a Korean individual Nature 2009 6 Ancient human genome sequence of an extinct Palaeo-Eskimo Nature, 463 2010, 7 The sequence and de novo assembly of the giant panda genome, Nature, 463 2010, 8 A map of human genome variation from population-scale sequencing, Nature, 467 2010

The next generation technology is known to take about two weeks to sequence the genome of a human (x30). Currently, more than 20,000 sequencers are reported worldwide, and about 500 billion won has recently been invested in the major development companies of the third-generation sequencer (Ion Torrent: 2.5 generation, third generation of Pacific BioScience) .

In addition, more than KRW 10 trillion has been invested in the relevant fields around the world and development is proceeding. With this trend, it is expected that the total sequencing cost of one person will be reduced to about $ 1,000 in the next two to three years. The most highly available and readily available technologies based on these next generation technologies are expected to be clinical genomics, pharmaco-genomics and translational medicine.

The clinical genome can be used to diagnose genomic information, which is NGS sequencing information, and it becomes clinical genome and customized medicine when combined with drug metabolism related technology including drug homogeneity test. In addition, genomics has been standardized at a high level. In particular, the Korea National Institute of Pharmacogenomics has documented a large number of evidence including more than 650 drug-genome-related SCI papers and 88 patents at the end of the nine-year project in March 2011 He contributed to the International Pharmacogenomics Society.

As shown in Fig. 1, Korea now has about 3,000 full-length exome sequencing programs, and the Seoul National University Asian Genome Center has 100 generators Sequencing is being performed. Also, internationally, genome sequencing for 1,000 people ended in 2010, and sequencing is now being performed worldwide, which is more than ten times more than the results achieved so far.

From a technical point of view, current genome (base polymorphism chip or next-generation sequencing technology) data-based biomarker search and discovery uses SNP (single nucleotide polymorphism) method. And the method of calculating this base polymorphism is called SNP calling.

And, as shown in FIGS. 2A and 2B, a base polymorphism calculation calculates base polymorphism (SNP) by applying statistics based on an allelic gene and performing base polymorphism definition.

Thus, biomarker discovery and detection techniques are used in association studies and linkage studies using normal and patient population polymorphism information.

On the other hand, when image information of next generation sequencing (NGS) and base polymorphism (SNP) chip data is processed, information such as allele difference, signal intensity, allelic imbalance and quality score . Based on this continuous variable data, variant calling is performed, and markers giving difference between normal and disease are selected with classified information (SNV, CNV, allele direction and INDEL).

Here, data classified as genotype corresponds to a categorical variable which is a discontinuous variable. Because these categorical variables are much more information lost than continuous variables, they conduct disease association and linkage studies due to rare alleles such as cancer, rare diseases and chronic diseases. The biomarker detection and excitation power tends to decrease.

Generally, in order to define a SNPV (SNV), the next generation sequencing (NGS) or nucleotide polymorphism - chip data is carried out by chemical methods on a large number of oligonucleotide chips, It is a common practice to quantify the intensity of a signal that indicates whether or not DNA fragments are tightly bound to each other after hybridization with DNA fragments attached to the chip. For SNPChip data, quantified signal intensity values are expressed as hundreds to thousands of values per base.

As shown in FIG. 3, in the case of SNP chips of illumina and affymetrix which are generally used, 1 million base polymorphism is integrated at one time.

Thus, producing a signal value (about 1,000) at 1 million alleles results in a value of 1M * 1,000, or 1 billion per genome, and processing 10,000 in this way results in a value of 100 trillion . Therefore, the data size is about 5-10TB.

And scientists digging biomarkers use only computed base polymorphism information (SNV) (about 10 GB) after processing using the SNV calling method.

On the other hand, Korean Patent Registration No. 10-0996443 discloses a method for processing a highly integrated gene database.

However, the above-described prior art has the following problems.

That is, when calculating the candidate group of the disease displacement marker candidate using the base polymorphism definition, since the data amount of the disease displacement marker candidate group is increased and there is no discrimination between the detection probability of the allele gene marker candidate candidate group, There is a problem in that the process of finding the data is complicated.

Korean Patent Registration No. 10-0996443

Disclosure of Invention Technical Problem [8] Accordingly, the present invention has been made keeping in mind the above problems occurring in the prior art, and it is an object of the present invention to provide a method of detecting a disease marker, The present invention provides a system and a method for detecting a disease variation marker of NGS data by an inverse classification method based on allele gene depth, signal intensity and quality score.

According to an aspect of the present invention for achieving the above object, the present invention provides a method for rapidly calculating a value for a quick calculation by using an allelotype depth, an imbalance, a signal strength value, a quality score, It is possible to extract high-sensitivity biomarkers based on the method of handling big data and reduce the burden of handling large-scale data by extracting only the raw data and erasing the raw data.

FIG. 4 shows a comparison between a method of discovering and utilizing a disease mutation marker by a conventional method and a method of diagnosing and diagnosing a disease mutation according to the present invention.

Table 2 below also shows SNP-based marker calculations using allelotype depth, unbalance, signal intensity value, and quality score. A simple example based on example sentences and allele depth is used to show that between two genomes with rare disease and three genomes with normal disease Of alleles of six pairs of genomes.

Here, GATK is a tool for SNV calling made by the Broad Institute of Harvard / MIT, and ADISCAN is a development tool according to the present invention.

Allele depth, signal intensity, and QS-based analysis example Pair dielectric GATK ADISCAN Pair1 (normal / disease) 202,421 44,852 Pair2 (normal / disease) 208,598 32,657 Pair3 (normal / disease) 197,496 24,793 Pair4 (normal / disease) 225,225 65,536 Pair5 (normal / disease) 239,990 48,878 Pair6 (normal / disease) 224,691 38,038 6pair intersection 31,145 5,153 GATK and ADISCAN common 4,892 4,892 * False Positive Including 25,992 0 * True Positive Missing 291 0

* Verification method: manual inspection

GATK is 31,145 and ADISCAN is 5,153 when the intersection of 6 pairs is obtained. In addition, 31,145 includes 31,145-5,153 = 25,992 false positives and includes 291 true positive losses in total. Therefore, both the sensitivity and the specificity of ADISCAN are superior to those of the GATK polymorphism definition (SNP calling) method. Accuracy and precision together make the sensitivity about (25,992 + 291) to 0, or 26,000 times more accurate. The accuracy here refers to the accuracy based on numerical values on a computer prior to experimental verification.

Despite the difficulties in handling biovig data, research and forecasting genome data and diagnostic tests using large-scale allele gene depth information based on the technology of RVR engine (Korean Patent No. 10-0880531) We have succeeded in developing a method for searching the gene traits consistently and with high sensitivity, and there are two different applications as follows.

(1) Detection of mutations in known genetic alleles of individual genomes.

(2) A new disease-specific genetic trait discovery method for cancer, rare disease and chronic disease genomes.

1. Identification of genetic variants of known genetic variants

The DNA of the human genome has 3 billion nucleotide sequences. Among them, monogenic disease and polygenic disease are divided into major disease causative mutations (single gene-based disease, multiple gene mutation, etc.) ) Are known to be about 10,000. And 100,000 mutations (complex diseases) that are indirectly related to the disease are also known.

Because most human DNA (except mitochondria and Y chromosomes) has two alleles (A: B) at all base positions, all of these genetic traits are alleles and haploids, a group of alleles, Or a haplogroup or genoset composed of a nucleotide sequence. Therefore, it is important to know the specific genetic polymorphism information of a specific gene among three billion genetic traits, and whether there is a disease trait among two allelic traits (A: B).

It is impossible to know the extent to which existing genetic traits are applied in Korean ethnicity because the genetic trait mutations that cause the known diseases are based on various races. Sensitivity is also an important criterion for biomarker detection due to rare alleles such as cancer and rare diseases.

Thus, using this outbreak, it is now easier to compute existing known base polymorphism and mutation data in a population of known populations in the Korean population, of which two of the allelic traits are disease-genotypic and which are genotypic.

FIG. 5 shows that a mutation detected in an individual is detected, and that a significant difference among the detected mutations can be confirmed in the overall result (higher significance graph) and individual result (+).

2. New disease-specific genetic traits of cancer and rare disease genomes

There are about 10,000 known biomarkers for the rare diseases, chronic diseases and cancers that are currently known, but they are not less than one-hundredth of the total biomarker pool. With the help of next-generation sequencing technology, a myriad of new biomarkers . However, sensitivity is also the most important criterion for biomarker discovery due to rare alleles such as cancer and rare diseases.

That is, if family history is known for rare diseases, it is necessary to apply biomarker search based on allele depth.

If a marker is discovered based on a normal SNP, more than 200,000 detection steps in the first step and about 30,000 in the second step are generated. Since the score is not displayed in the 30,000 generated markers, Lt; / RTI >

That is, in the case of cancer, it is necessary to search for a pair of full-length genomes based on the allele-type gene depth. Also, when cancer recurs or cancer tolerance develops, it can be measured with high sensitivity based on the allele depth of this allele.

In Fig. 6, the I-2 gene is a well-known biomarker, and when sorted by this method, seven high-sensitivity results are detected. If a marker is discovered based on normal SNP, it will have a mutation that gives 100 to 200 times more significance than the above-mentioned sorted result. Further, there is a further need for a procedure for verifying with much time using it. However, using the method of the present invention, verification can be performed from above because the overall result is scoring. In addition, using the present invention, additional validation results can be generated using the allele-specific depth DB of a non-cancer-affected general person, such as the above-mentioned rare disease.

FIG. 7 shows an example of the execution of an allele marker in an actual multiple leukemia patient.

3. Possibility of specification of technology due to transition from next generation to third generation technology

A key feature of next-generation sequencing technology is the detection by chemical hybridization of chip or bead-based DNA Oligo fragments. However, the third generation technology that quantifies the flowtone that occurs when DNA is synthesized and the base that discriminates subtle changes when the DNA is synthesized by reading the base by light, and the RainDanse technology that uses aggregation of oil droplets Will be developed as a step of generating a Haploid sequence, which is still useful as a vehicle body, but still involves copy number variation of DNA, which is the cause of the signal intensity value of the next generation sequencing data, It is still useful in the problem of finding disease markers using information of good or not of sequencing due to the conservation of protein and it is highly likely to develop into next generation technology body forever.

In addition, the application of the present patent can detect disease markers that give differences in normal and disease due to alleles and signal intensity values, such as detection methods for ChipSeq, RNAseq, and DNA METseq, which are other areas of NGS.

DISCLOSURE OF THE INVENTION According to the present invention, there is provided a system and method for detecting a disease variation marker of NGS data based on allele depth, signal strength, and quality score-based inverse classification method, The present invention provides a method of detecting an actual disease displacement marker from the candidate group by assigning ranking according to the possibility of detecting a disease displacement marker on the data of the disease displacement marker candidate group when calculating the candidate group of the disease displacement marker of the allele have.

In addition, this patent technology, PMAP (individual genome map) technology and patient-specific genotyping technology based on the individual can make a great contribution to the field of personalized medicine.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic diagram illustrating a custom medical genetic service service in progress;
Figure 2A is an example of SNPchip signal intensity values for base polymorphism calculation.
Fig. 2B is an illustration showing the NGS allele-type gene depth for calculation of base polymorphism. Fig.
Figure 3 is an illustration showing a human genome disease allelic variant discovery and search schema.
FIG. 4 is a flow chart comparing the process of discovering an allelic variation according to the prior art and the present invention. FIG.
Figure 5 is an illustration showing an example of searching for known disease alleles of individual genomes according to the present invention.
6 is an illustration showing an example of discovering a disease allele marker in a rare disease according to the present invention.
FIG. 7 is an illustration showing an example in which a disease allele marker is excavated from a patient with multiple leukemia according to the present invention. FIG.
FIG. 8 is an exemplary diagram showing allele difference tangent differences according to the present invention. FIG.
FIG. 9 is an exemplary diagram showing differences in geometric alleles according to the present invention. FIG.
10 is an illustration showing an example of searching for a personal genomic disease allele marker by the present invention.
11 is an illustration showing an example of discovering a rare disease family data new disease marker according to the present invention.
Figure 12 is an illustration showing an example of discovering a female pair-data new disease marker by the present invention.
Figure 13 is an illustration showing an example of discovery of an arm-pair data drug resistance allele by the present invention.
14 is an illustration showing an example of discovering a chronic disease allele new marker by the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a disease variance marker detection system and method according to a specific embodiment of the present invention will be described in detail with reference to the accompanying drawings.

First, in describing the present invention, terms used will be defined as follows.

- Single nucleotide polymorphism (SNP): Genetic trait information that can have 2 to 5% different alleles in a population.

- Next Generation Sequencing (NGS): Techniques for fragmenting a full-length genome in a chip-based and PCR-based paired end format and performing sequencing at high speed based on chemical hybridization of the fragment .

- Whole genome sequencing (WGS): A method of sequencing whole genomes by next generation sequencing methods (NGS) to read the human genome in multiples of 10X, 20X, and 40X formats.

- Whole exome sequencing (WES): means that only the gene region involved in protein production is sequenced among the above WGS.

- TS (Target sequencing): means sequencing only the gene region involved in the target protein production among the above WGS. Thus, the data size of WGS> WES> TS is generated. However, because of its small size, it has the advantage of being able to sequence many samples.

- METSEq: Sequencing technology for DNA methylation measurement of genes

- RNAseq: gene expression, ie, sequencing technology for DNA transcriptome

- Microarray: A chip-based technology for gene expression, DNA transcriptome, similar to RNAseq, but a traditional chip-based technology rather than sequencing.

- SNPchip: Affymetrix and Illumina are at the center, and now chip technology means that 1 million SNP information can be accumulated in one chip-array.

- Cell or tissue population: There are many different cells in the human body. Cells such as leukocytes, eye cells, brain cells, hepatocytes, and cancer cells are mostly separated separately, but they are not separated when performing sequencing. The DNA can be extracted without any change. Especially when cancer cells are extracted, normal cells are extracted together. To express this phenomenon, each group of different cells is referred to as a cell group.

- human population: means a general population with different genetic traits such as Korean, Japanese, Chinese, Black, and Kokesian.

- Next generation sequencing: refers to sequencing technology such as illumina, Roche, and Toronto, and broadly includes technologies such as third generation technologies Pacificbio, RainDance, and fourth generation technologies.

The difference between these technologies is the next generation technology based on the chemical reaction of oligo on a chip or bead basis and the 3rd generation or the 4th generation when the speed and sequencing technology are dramatically improved in the way of decoding DNA .

- Allele depth: When sequencing is performed, DNA sequencing is applied to the standard genome. All of these genes are called total depth. In total depth, allele A and relative allele B are included. The sum of these two different alleles is called depth and the individual early is called the standard, relative allele.

Signal intensity: When a DNA fragment such as an oligo is attached to a piece of DNA attached to a chip, the color and intensity are measured according to the characteristics of the binding and the base.

- Amino acid mutation: As a result of base polymorphism, a variant of a wild type amino acid is called a mutation.

- Sequencing folds: It means the number of times to perform sequencing. Sequencing is generally done at 10x, but when 20x is used, the result is usually. When sequencing at 40x multiple times, Is possible.

- Locus: means the position of one base on the chromosome.

Linkage disequilibrium (LD): A method of testing whether two loci on multiple chromosomes have a linkage disequilibrium.

- structural variation (SV): A variation in a large unit (dna segment) of a chromosome resulting from insertion, inversion, or translocation in a variation.

- Directional SNV: Two alleles have different orientations. The homozygous locus of one normal genome is a unique case where the locus of another genome is an alternative homo, which is seen in different races. However, screening can be done when these mutations are found in cancer cells or in specific diseases. In addition, there are cases where two hetero atoms have different orientations, but the cause is not yet known.

- Locust epistasis: a phenomenon where a variation in a particular locus is involved in the regulation of a nearby gene or a remote gene (eg, alternative splicing).

- Locus functional information: means that the locus on each chromosome is functionally known (eg, CNV, INDEL, Alternative splicing, promoter, etc.)

-SAMTOOLS: Primitive sort information manipulation and mutation definition engine

-mpileup: One of the functions used by SAMTOOLS, it is possible to arrange several BAM files on the x to y axis on the y to x axis

-BAM: Binary sequencing Alignment Map file,

-SAM: Sequencing Alignment Map file,

Hereinafter, a disease mutation marker detection system and method of NGS data according to the allele gene depth, signal intensity score, and locus function information-based inverse classification method according to the present invention will be described in detail with reference to the accompanying drawings .

FIG. 8 is an illustration showing the allele gene depth difference according to the present invention, FIG. 9 is a diagram illustrating the difference of geometric alleles according to the present invention, and FIG. 11 is a diagram illustrating an example of discovering a new disease marker for rare disease family data according to the present invention, and Fig. 12 is a diagram showing an example of searching for a cancer pair- FIG. 13 is an exemplary diagram illustrating an example of discovery of a cancer-pair data drug resistance allele by the present invention, and FIG. An example of discovering a new marker of a chronic disease allele is shown.

First, a first embodiment of the present invention will be described.

Example 1: The first embodiment of the present invention is a method for screening a pair-dielectric specific variation marker.

The ADISCAN (allelic depth and imbalance scanning) according to the present invention uses the independent or weighted mixture of seven analytical methods to screen for markers giving all differences in normal and disease genomes.

We define each algorithm as t, s, d, g, c, r, and e as a function of t, w, Method.

C: statistical allele difference r: allelic imbalance ratio, and e: locus function. E: allelic difference, t: allele depth product tangent difference, s: allelic ridge, d: absolute difference in allele, g: geometric allele difference, Information. w1, w2, w3, w4, w5 and w6 are empirical (or user operable) weight values. And e is a weight value for the function information of a specific locus in each genome. That is, it means table lookup data such as epistasis, multiple linkage disequilibrium (LD), and known functional information.

The details of each of these are as follows.

1) Allele depth tangent difference

The alleles of the normal genome at the locus of the entire genome are called A and B, and alleles of the disease genome are called C and D. In the case of multiple genomes, the A, B, C and D values (A, B) and y = adi (C, D), respectively, when the two alleles have depths (A, B) or (C, D). Here, adi () is a continuous variable (= x) starting from 0 that transforms various alleles (A, B) or relative (disease) alleles (C, D) It is a continuous variable of constant form (= y). The largest value of A, B, C, or D at the entire locus of two dielectrics is defined as max. However, if max is greater than the absolute value E, then max is defined as E and the allele depth tangent difference is

t = 1 / tan (y / x) * 1 / tan (max - x / max - y)

Here, the t value is obtained by using the allelic depth tangent difference (t) as three orders of x and y (t1, t2, t3) in the two-dimensional coordinate system under the x and y axes, When sorted, t1 is ranked 1, t3 is ranked 2, and t2 is ranked 3. In this way, the locus that gives the greatest difference between the two normal and disease genomes can be aligned (Figure 8).

2) Allele rats

The alleles of the normal genome at the locus of the entire genome are referred to as A and B and the alleles of the disease genome are defined as C and D. In the case of multiple genomes, A, B, C and D Assuming that the values mean the mean, the allele ranks (s) can use various squared powers such as 2, 3, 4, 5, 6, and N squared. In particular, the squared power that is well suited to the model is six power squared, as follows. x = B 6 / A, A> = B and A> = 1, and y = D 6 / C where C> = D and D> = 1. Also, the denominator is a large number in the depth of two alleles, and the numerator is a small number. There are a few other ways to determine if the allele depth is 0, or to double the relative allele depth of 0 to balance.

s = | log (x) - log (y) |

The feature of this algorithm is that it expresses the logarithm of the 6 squared difference of two different alleles, and skewed distribution from 0 to infinity.

In addition, we use x1 = log (B ^ 6 / A) and x2 = log (B ^ 6) as a method of performing variant calling using the logarithm of various squared powers (B / A ^ 6), y1 = log (D ^ 6 / C), and y2 = log (D / C ^ 6). And this score means homozygote when x1 is minus (-) and heterozygote if x1 is positive (+). In the same way, when y1 becomes minus (-), it means homozygote, and when y1 is positive (+), it means heterozygote. And, even if x1 and y1 are positive ("+"), if the score is between 0 and 1, an area where the variation definition can not be determined, ie, a twilight zone or a partially defined variant calling ), And may make final decisions using other locus function information, association imbalance (LD) information, or sample population frequency information. Also, x2 and y2 are utilized as information for visualization for directionality and balance of the final homo and hetero.

3) Difference in absolute value of allele

The alleles of the normal genome at the locus of the entire genome are referred to as A and B and the alleles of the disease genome are defined as C and D. In the case of multiple genomes, A, B, C and D If the value means the mean value, the difference (d) of the absolute value of the allele is as follows.

d = sqrt (| A-B | * | C-D |)

The feature of this algorithm is to maximize the difference of alleles. Also, because the d value includes the simplest formulas, it can be useful in terms of verifying the inconsistencies arising from alignment by complex formulas.

4) Geometric allele difference (g = geometry-based difference)

As shown in FIG. 9, the geometric allele difference is represented by A and B as alleles of the normal genome at the specific locus of the entire genome, and C and D as the alleles of the disease genome, If the A, B, C, and D values of the normal and disease mean mean values, the transformation for the geometric allele difference calculation is as follows.

small = min (A, B, C, D);

A = A - small +1;

B = B - small +1;

C = C - small +1;

D = D - small +1;

x = (A + B + C + D) / 4;

AB = sqrt ((0-A) * (0-A) + (1 * x-B) * (1 * x-B));

BC = sqrt ((1 * x-B) * (1 * x-B) + (2 * x-C) * (2 * x-C));

CD = sqrt ((2 * x-B) * (2 * x-B) + (3 * x-C) * (3 * x-C));

DA = sqrt ((3 * x-B) * (3 * x-B) + (0-A) * (0-A));

a = AB; b = BC; c = CD; d = DA;

s = (a + b + c + d) / 2.

The difference (g) between the normal and disease geometric areas using the transformed new information is as follows.

(s-d))) * 1 / sqrt (small) g = log (sqrt ((s-

Here, AB, BC, CD, and DA mean the distance of two alleles. The final g (= geometry-based difference) refers to the square Brahmagupta area connected by four lines. Then, 1 / sqrt (small) is divided by the smallest value of each allele and normalized. If this is done with non-standardized values, The allele ratio value of the disease is 250: 250 vs.. 100: 250 and 250: 250. It is difficult to distinguish the difference of 1: 250.

In other words, the former has an allele depth of 150, while the latter has an allele depth of 250, but the geometric difference algorithm is calculated as 250 to 250, so standardization is needed. Also, the g value can be useful in terms of verifying the inconsistencies in complex mathematical sorting because it includes two alleles and a formula for the simplest area with x and y axes.

5) Statistical allele difference (c = 2x2 contingency chi-square)

The alleles of the normal genome at the locus of the entire genome are called A and B, and alleles of the disease genome are called C and D. In the case of multiple genomes, the A, B, C and D values Is the mean value, the 2x2 contiguous Kaisier statistic (c) of the normal and disease genomes is:

c = chi-square (A, B, C, D);

In addition, the c value is a measure of sample contamination by the fisher exact test to the extent to which the degree of sample contamination of the two alleles is proportional to the ratio of the most ideal (or hypothetical) Available.

6) Allele imbalance ratio (r = log2 allele imbalance ratio)

If the alleles of the normal genome at the locus of the entire genome are called A and B and the alleles of the disease genome are C and D, the ratio of allelic imbalance values of normal and disease (r) is as follows:

ratio = max (A + B, C + D) / min (A + B, C + D);

r = log2 (ratio);

In addition, the r value is used as a means of measuring copy number variation (CNV) and loss of heterozygosity (LOH) since it can detect the degree of increase / decrease of two alleles.

7) Locus function weight (e)

In the specific locus position of the entire genome, e is a weight value for the functional information of a particular locus in each genome. That is, it refers to table lookup data such as epistasis, linkage disequilibrium (LD), and known functional information.

Next, a second embodiment of the present invention will be described.

Example 2: The second embodiment of the present invention is a method of analyzing difference by allele-depth-based reverse copy number (CNV), SNV, INDEL, directional SNV, and SV (structural variation) method.

When the ADISCAN according to the present invention is applied in the manner summarized in Example 1 as described above, the difference between two loci (positions on the chromosome) is calculated in the form of a score in which allele depth and intensity are different.

And the score has 7 different meanings.

Here, the first is a method of weighting when one dielectric is homo and the other is heterogeneous, the second is base polymorphism (or mutation difference), the third is CNV difference, the fourth is INDEL difference, The fifth searches for all structural alleles with SV (Structural variation) differences, the sixth the directional SNV, and the seventh searches the given locus function weights (eg, topology, association disequilibrium (LD) It is expressed in the form of the sum of the scores of the seven methods searched and the score of one representative among 7 kinds.

Therefore, the total score is a mixture of scores from seven different factors. The first objective is to extract the marker of disease mutation that differentiates between the normal and the disease genome. Therefore, all the genes and locus portions (SNV, CNV, INDEL, Directional SNV and SV) and apply the final classification result (SNV, CNV, INDEL, directional SNV and SV) , And iii) thereafter reverse-transforms the five kinds of information into variant calling according to the original purpose.

1) Selection of disease mutation markers by reverse classification-based polymorphism

The results of ADISCAN are highly rated when the two alleles are unusually different. Therefore, SNV calling for the selected alleles can be used to calculate the nucleotide polymorphism (or mutation) difference. With this, we can explain that the two Zino types are different. In other words, it is a method to calculate the difference first and to perform the base polymorphism definition on the meaningful one.

2) Selection of disease variant markers by reverse classification based CNV

The results of ADISCAN are highly scored when the two alleles are different, which is caused by the difference in the number of DNA fragments (CNV), especially when the signal intensity value is high. Thus, the CNV difference can be calculated by performing a CNV calling on the selected alleles. In other words, the marker that gives allelic difference is first calculated and CNV definition is performed on the meaningful one.

3) Reverse sorting based on INDEL (Insertion-Deletion)

The results of ADISCAN are highly scored when the two alleles are different from each other, especially when the difference in alleles is measured to be high, which is caused by differences in the insertion and deletion of DNA fragments. Thus, the difference in INDEL can be calculated by performing an INDEL calling on the selected alleles. In other words, the marker that gives allele difference is calculated first, and the INDEL definition is performed for the meaningful one.

4) Selection of disease mutation markers using reverse classification-based allele SV

The results of ADISCAN are highly scored when the two alleles are different. Here, especially when the difference in alleles is measured to be high, the long insertion and long deletion of the DNA fragment, This is caused by differences in translocation. Therefore, the SV difference can be calculated by performing SV calling on the selected alleles. That is, it is a method to calculate the marker giving the allele difference first and to perform the SV definition on the meaningful one.

5) Classification based on alleles Segregation of disease markers using directional SNVs

Two pairs of allelic information with different orientations are defined as Hetero in many cases and are not caught by other genotypes between disease and normal. In particular, homozygosity or heterozygosity can not be clearly distinguished in calculating base polymorphisms using a pair of alleles. In this case, both the normal and the patient's control group can be calculated to give distinctly different alleles. In this way, when there is only one genome, unambiguous information can be obtained by comparing the two genomes with each other, It is first to select genes using the difference, and if they are selected as Homo and Hetero after selection, they can be identified as markers that give a difference between normal / diseased cells that are not homo or heterozygous.

Next, a third embodiment of the present invention will be described.

Third Embodiment: A third embodiment of the present invention relates to a method for generating a normal / patient integrated DB for verification.

The third embodiment of the present invention is based on the finding that allelic depth, signal intensity, marker call rate (MCR) in multiple samples, and linkage disequilibrium on chromosomes in a large- A linkage disequilibrium (LD) score, and a quality score, to generate a DB. Assuming that there are 500,000 locus of 100 samples, DB can be generated as follows.

Step 1: Most SNV calling algorithms generate VCF (variant call format) together with SNV information as an output file. However, in the third embodiment due to the incidence, all parameters are set to "0 " so that the allele pairs Allele A and Allele B and their Quality Score, LD, MCR, and QS (QS1: phred quality score, QS2: mapping quality score). Or the mpileup (function to sort multiple sort files from x to y axis and from y to x) functions in samtools (native sort information manipulation and mutation definition engine) functions. Can be calculated and extracted.

Step 2: Combine the extracted values into 6 values per sample and arrange them as shown below.

1, QS1 [1], QS2 [0], LD [0], MCR [0], AlleleA [1], alleleB [ ], LD [1], MCR [1], ... AlleleA [N], alleleB [N], QS1 [N], QS2 [N], LD [N] and MCR [N].

Where N is 100. Thus, a total of 6 * 100 + 1 (Chr_Position) = 601 columns is generated. The file thus arranged is indexed for all the numbers in the RVR file (refer to Korean Patent Registration No. 10-0996443), and the necessary column and line information can be automatically extracted. In the above example, the size of the generated DB is 601 (column) x 500,000 (line).

Step 3: If the disease variability markers generated by the first and second steps are X, the X markers are set to X for the normal / disease proof set generated through the first and second steps, Extract from DB and use for verification purpose.

In this way, a DB for normal control and DB for case disease can be created.

Next, a fourth embodiment of the present invention will be described.

Example 4: A fourth embodiment of the present invention is a method for searching for disease-related alleles in a personal genome.

In this case, the terms used are defined as follows.

- Individual genome: refers to the target genome to search for a gene that only gives a disease or phenotype without knowing whether it is normal or disease.

- AD: allele depth information (two alleles)

- QS: quality score

- N: Number of AD values of normal dielectric

- D: number of AD values of the disease genome

- X: Symbol indicating no difference in normal and disease AD

- Y: Symbol indicating difference in normal and disease AD

- ADISCAN: package variation information to determine the difference between normal and disease AD

The ADISCAN-based individual genome search method for disease alleles requires a database of alleles-specific verified disease allele genes. This is because approximately 10,000 known mutational information that is already known does not apply to all races.

Therefore, a diagnostic test is required to be performed using a verified biomarker that requires specific tests for individual races. In this embodiment, the ADISCAN-based verification method is applied.

There are currently 1000G (1000 genomes), KARE data (10,000 genomes) and PGM21 integrated data (3000 genomes) from the Next Generation Custom Medical Genome Corporation. A validation database is needed to validate ADISCAN results, such as aggregated population genomic data. This means that if one person performs NGS-based full-length exome sequencing (WES), the entire exome is about 50 MB in length, which means finding a known mutation in it and predicting the type of disease allele at the mutation location .

For example, if a healthy person performs the above tests, he or she can search for mutations in rare diseases that are activated at a particular age, including Mendelian, which causes illness.

As shown in FIG. 10, the fourth embodiment of the present invention can be performed through the following steps.

Step 1: Mapping to a region calculated at least 20% (or lower or higher) of the disease allele in a full-length genome population genome (1000genome, KARE, PGM21, etc.) including Chinese and Japanese similar to Koreans and Koreans We construct a DB that calculates disease markers including all the disease alleles.

If an international service is performed, the region that meets the above criteria for the disease allele of each country is selected, and the DB based on the pre-calculation work is used.

Step 2: Sequencing the individual genome using WGS, WES, or TS method.

Step 3: Extract all allele depth information at the position containing the disease allele in the sequenced genomic data.

Step 4: Use ADISCAN to perform the calculation of disease risk and ADISCAN score using the results of the second and third steps as input files.

Step 5: Using the DB information of the first step, the difference between the personal genome of the second step and the disease allele gene location information of the comparative population is calculated, N difference results. Calculate the significance number (X) and calculate the arithmetic success rate and sort from the highest to the lowest.

Rank 1, Allelee, Allele B, X * 100 / N, Sum (ADISCAN (i) * X (i)

Arithmetic success rate = X * 100 / N;

Arithmetic success rate score sum = Sum (ADISCAN (i) * X (i));

Using the above success rate and scoring by weight, generate a report form containing reference materials that can be read and evaluated by the user (physician or genetic information specialist) in the order of ranking and sorting. Here, X denotes a rank score, and N denotes the number of samples.

Next, a fifth embodiment of the present invention will be described.

Example 5: A fifth embodiment of the present invention relates to a method for discovering a new disease allele in a rare disease / cancer pair genome, as shown in Fig.

Based on the ADISCAN-based rare disease and cancer, a novel disease marker can be used to identify alleles of rare diseases and cancers in the family, and can be performed in the same way using a pair of samples containing their normal cells and cancer cells, This is possible.

The discovery of such new disease alleles can be performed through the following steps.

Step 1: Perform sequencing of normal and patient genomes by WGS, WES or TS method.

In this case, the patient dielectric refers to a genome of a rare disease patient or a genome of a cancer cell, and the normal genome is a genome of a genome or a cancer patient who has not developed the rare disease among families of a rare disease patient The genome of a family of unexposed cancer patients).

Step 2: In the sequenced data, the normal genomic and disease genome pairs are calculated using ADISCAN.

For example, normal (grandmother) vs. Disease (father), normal (grandmother) etc. Disease (son 1), normal (grandmother) vs. Disease (son 2), normal (aunt) vs. others. Disease (father), normal (aunt) etc. Disease (son 1), normal (aunt) vs. others. Six pairs of disease (Son 2) produce a mutation giving significance using ADISCAN.

Step 3: Final selection of all overlapping mutations. Thus, in the case of family data, the father-son-son-2 is likely to be caused by the same disease allelic gene, so all six pairs simultaneously calculate significant disease alleles.

Step 4: Utilizing the verified population-specific disease allele genetic trait database set in the first step and the second step set in the individual genome disease allele genetic trait search method. In the third step, 4 to 5 thousand candidate markers are selected from 50 million total data. When these candidate markers are compared with the population-specific disease allele genetic trait DB using ADISCAN, a higher accuracy biomarker target Can be selected.

Step 5: After the fourth step, the final marker candidates are determined in descending order of score, and CNV (copy number variance), bio map, tin work, and protein simulation are performed.

Next, a sixth embodiment of the present invention will be described.

Example 6: A sixth embodiment of the present invention relates to a method for locating a cancer resistant cause marker in a cancer resistant genome as shown in FIG.

The sixth embodiment is performed by the first to third steps in the same manner as the disease allele discovery in the rare disease / cancer genome of the fifth embodiment.

Hereinafter, an example of a marker discovery method for leukemia will be described as an example.

Among the samples, ID_G1, G2, G3, and G4 (Total 4 cases) have a special subtype of APL among leukemia (AML). The genomes of these patients have a t (15; 17) translocation and the gene has a PML / RARA gene rearrangement. In other words, they are already known as oncogenic drivers.

Therefore, the subtypes will use a target treatment targeting PML / RARA, which is mostly heard but not heard in some cases. G1 patient is the corresponding patient. You need to find a variation in G1 that is not in the rest. I suspect that the mutation may be related to refusal of the target treatment.

Since the subtypes are homogenous, the cancer mutations will not vary and are expected to be relatively identical. Thus, we look for a variation in G1 that is not in the rest. This variation is not of germline but of somatic.

As shown in FIG. 12, all of G1, G2, G3 and G4 in the above case are pair-samples. That is, G1 comprises two (pair) dielectrics of normal and cancerous tissue.

Step 1: Perform ADISCAN on each pair-dielectrics of G1, G2, ... G4 and select candidates.

Step 2: Select mutations that are specific only to G1 among all the significance-giving results.

Step 3: Perform a protein simulation to determine if the mutation is responsible for the drug resistance of G1.

Next, a seventh embodiment of the present invention will be described.

Example 7: A seventh embodiment of the present invention relates to a method for excavating a marker in a chronic disease genome, as shown in Fig.

The normal group AD information, N (s), disease group AD information, and D (s) are compared in a pairedwise manner. N1 x D (s), N2 x D (s), ... Nn x D (s). Then, N1.Y, N2.Y, and Nn.Y are calculated. Here, N (s) .Y means a marker list giving significance. Then, all the scores of N (s) .Y are sorted and displayed.

It is to be understood that the invention is not limited to the disclosed embodiment, but is capable of many modifications and variations within the scope of the appended claims. It is self-evident.

The present invention relates to a comparative analysis method using a continuous variable itself used to define a base polymorphism in order to improve biomarker searching and digging power. According to the present invention, sensitivity of searching for genome biomarkers is improved Which is 10 ~ 10,000 times better than the conventional method.

Claims (21)

Comparing one or more normal genomic DNA polymorphism data to one or more disease genomic polymorphism data;
Calculating a priority of a disease marker based on a predetermined classification criterion; And
And outputting the disease marker sorted by the priority order:
The sorting criterion may be,
Between the normal genome and the allele of the whole vaginal yeast oil,
(Absolute difference in absolute value of allele), geometric difference (geometric allele difference), statistical difference between tangent values of depth (tangent difference of allele), difference of squared value A difference value (statistical allele difference), an unbalance ratio value (allele imbalance ratio), or a locus function:
The priority order may be,
Is calculated as the sum of at least one of the weights of the variables t, s, d, g, c, r or e:
Where the variable t is a variable due to allele difference tangent difference;
The variable s is a variable representing the allelic squared ranks;
Wherein the variable d is a variable indicating an absolute difference in alleles;
Wherein the variable g is a variable representing a geometric allele difference;
Wherein the variable c is a variable indicative of a statistical allele difference;
The variable r is a variable representing an allelic imbalance ratio;
Wherein the variable e is a weighted score for the locus function.
delete The method according to claim 1,
The variable " t "
is calculated by t = [1 / tan (y / x)] * [1 / tan (max-x / max-y)
Here, the adi () function is a function that transforms an allele to a constant of various types;
x = adi (A, B) and y = adi (C, D);
A and B refer to alleles of the normal genome at the specific locus position of the entire genome;
C and D are alleles of the disease genome at the locus of the entire genome;
Wherein the alleles of normal and disease are weighed against opposite concepts such as homo and heterozygous.
The method according to claim 1,
The allelic squared rider (s)
Wherein the absolute value of the log unit difference value of the two values calculated from the depths of the two different alleles is ranked as a score, thereby detecting a marker of disease variation in the next generation sequencing data.
5. The method of claim 4,
The allelic squared rider (s)
s = | log (x) - log (y) | Lt; / RTI >
The variables x and y,
x = B ^ N / A (A > = B and A > = 1);
y = D ^ N / C (C > = D and D > = 1);
A and B and C and D are variables that represent different allele depths;
Wherein N is an integer of 1 or more.
6. The method of claim 5,
Wherein the N is 6. ≪ RTI ID = 0.0 > 8. < / RTI >
5. The method of claim 4,
The allelic squared rider (s)
(Homo, hetero) variant of a homo and hetero form of minus ("-") or plus (+) forms generated from the logarithm of the allele. Detection of mutation markers.
8. The method of claim 7,
The allelic squared rider (s)
is calculated from the difference between log (B ^ 6 / A) or log (B / A ^ 6) and log (D ^ 6 / C) or log (D / C ^
A and B, and C and D are variables representing different allele gene depths, respectively, in the next generation sequencing data.
The method according to claim 1,
The variable " d "
d = sqrt (| AB | * | CD |):
Where A and B refer to alleles of the normal genome at the locus of the entire genome;
Wherein C and D are alleles of a disease genome at a specific locus of the entire genome.
The method according to claim 1,
The variable " g "
It is a variable that reflects the difference in the geometric area of the normal and diseased genomes,
calculated from g = log sqrt ((sa) * (sb) * (sc) * (sd))) * 1 / sqrt (small)
here,
small = min (A, B, C, D);
A '= A - small +1;
B '= B - small +1;
C '= C - small +1;
D '= D - small +1;
x = (A + B + C + D) / 4;
AB = sqrt ((0-A ') * (0-A') + (1 * x-B ') * (1 * x-B'));
BC = sqrt ((1 * x-B ') * (1 * x-B') + (2 * x-C ') * (2 * x-C'));
CD = sqrt ((2 * x-B ') * (2 * x-B') + (3 * x-C ') * (3 * x-C'));
DA = sqrt ((3 * x-B ') * (3 * x-B') + (0-A ') * (0-A'));
a = AB; b = BC; c = CD; d = DA;
s = (a + b + c + d) / 2 and:
A and B refer to the alleles of the normal genome at the specific locus position of the entire genome and C and D refer to alleles of the disease genome at the specific locus position of the entire genome and AB, BC, Wherein the CD and the DA each represent the distance between two alleles of the next generation sequencing data.
The method according to claim 1,
The variable c is a variable reflecting a 2x2 contingency chi-square statistical value between the allele of the normal genome and the allele of the disease genome;
The 2x2 continea-kinase statistic (c) of the normal and disease genomes,
c = chi-square (A, B, C, D);
A and B denote alleles of the normal genome at the specific locus of the entire genome and C and D denote alleles of the disease genome at the specific locus position of the entire genome. Detection method of disease variation markers in sequencing data.
The method according to claim 1,
The variable " r "
This variable reflects the ratio of the allelic imbalance of the normal and disease genomes,
r = log2 (ratio);
A and B are alleles of a normal genome at a specific locus position of the whole genome, and the ratio of the allele of the normal genome to the specific genome of the entire genome is R = max (A + B, C + D) / min C and D are alleles of disease genomes at specific locus loci of the entire genome.
13. The method according to any one of claims 1 to 12,
(A) extracting, for the detected disease mutation marker, a portion where a difference between two dielectrics occurs;
(B) classifying the difference of one of the base polymorphism, copy number variation, INDEL (Insertion-Deletion) or allelic directionality, and structural variation (SV) reverse to the difference occurrence part; And
(C) outputting a difference classification (variant calling) classified into a classification criterion of the step (B). ≪ Desc / Clms Page number 20 >
delete delete delete delete delete delete delete delete

Images