KR101816646B1 - A METHOD FOR PROCESSING DATA OF A COMPUTER FOR IDENTIFYING GENE-microRNA MODULE HAVING HIGH COREELATION WITH CANCER AND A METHOD OF SELECTING GENES AND microRNAs HAVING HIGH COREELATION WITH CANCER - Google Patents
A METHOD FOR PROCESSING DATA OF A COMPUTER FOR IDENTIFYING GENE-microRNA MODULE HAVING HIGH COREELATION WITH CANCER AND A METHOD OF SELECTING GENES AND microRNAs HAVING HIGH COREELATION WITH CANCER Download PDFInfo
- Publication number
- KR101816646B1 KR101816646B1 KR1020150110861A KR20150110861A KR101816646B1 KR 101816646 B1 KR101816646 B1 KR 101816646B1 KR 1020150110861 A KR1020150110861 A KR 1020150110861A KR 20150110861 A KR20150110861 A KR 20150110861A KR 101816646 B1 KR101816646 B1 KR 101816646B1
- Authority
- KR
- South Korea
- Prior art keywords
- gene
- cancer
- module
- genes
- microrna
- Prior art date
Links
Images
Classifications
-
- G06F19/18—
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12N—MICROORGANISMS OR ENZYMES; COMPOSITIONS THEREOF; PROPAGATING, PRESERVING, OR MAINTAINING MICROORGANISMS; MUTATION OR GENETIC ENGINEERING; CULTURE MEDIA
- C12N15/00—Mutation or genetic engineering; DNA or RNA concerning genetic engineering, vectors, e.g. plasmids, or their isolation, preparation or purification; Use of hosts therefor
- C12N15/09—Recombinant DNA-technology
- C12N15/11—DNA or RNA fragments; Modified forms thereof; Non-coding nucleic acids having a biological activity
- C12N15/113—Non-coding nucleic acids modulating the expression of genes, e.g. antisense oligonucleotides; Antisense DNA or RNA; Triplex- forming oligonucleotides; Catalytic nucleic acids, e.g. ribozymes; Nucleic acids used in co-suppression or gene silencing
-
- G06F19/20—
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Genetics & Genomics (AREA)
- Biomedical Technology (AREA)
- Chemical & Material Sciences (AREA)
- Molecular Biology (AREA)
- Organic Chemistry (AREA)
- Biotechnology (AREA)
- General Engineering & Computer Science (AREA)
- Zoology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Wood Science & Technology (AREA)
- Microbiology (AREA)
- Plant Pathology (AREA)
- Physics & Mathematics (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
Abstract
A method for identifying a gene-microRNA module highly related to cancer according to the present invention comprises: collecting and pretreating gene and microRNA expression data; Constructing a gene-cancer patient module utilizing bi-clustering; Expanding the gene into a gene-cancer patient module; Constructing a gene-microRNA module; And verifying the association with cancer.
Description
The present invention relates to an approach for gene-microRNA module identification that is highly relevant to cancer.
The microRNA is a short non-coding RNA consisting of about 22 nucleotide sequences and is known to function as a transcriptional regulator in the process of gene expression.
MicroRNAs play an important role in the onset and progression of various cancers by controlling genes. Since many microRNAs can regulate a large number of genes, the mechanism of regulation between microRNAs and genes is complex. In addition, even patients of the same type of cancer may exhibit other regulatory mechanisms.
On the other hand, in addition to direct gene regulation by microRNAs, the relationship between genes and microRNAs is very complex, since transcription factors may be involved in gene and microRNA regulation.
Because of this complexity, it is difficult to predict the association of cancer and cancer with the analysis of genes and microRNAs.
In addition, the majority of studies have identified genes that are directly regulated by microRNAs, and they are composed of a single module, so it is difficult to grasp the relationship between gene-microRNAs and to understand the mechanisms of various cancers .
The present invention seeks to identify the relationship between genes and microRNAs in cancer patients using expression data.
In detail, the present invention establishes a module consisting of a set of genes and microRNAs that are highly correlated in expression level, said module being comprised of genes and microRNAs associated with a particular cancer type, said module being capable of direct MicroRNA module identification that is highly correlated with cancer constituting both regulatory and indirect regulation via transcription factors.
A method for identifying a gene-microRNA module highly related to cancer according to the present invention comprises: collecting and pretreating gene and microRNA expression data; Constructing a gene-cancer patient module utilizing bi-clustering; Expanding the gene into a gene-cancer patient module; Constructing a gene-microRNA module; And verifying the association with cancer.
(1) when a microRNA directly regulates a gene, (2) when a microRNA controls a transcription factor and the transcription factor controls a gene, and (3) when a transcription factor binds a gene and a microRNA simultaneously And the like.
That is, the reliability of prediction of association with cancer is improved by reflecting various interactions into the module by the gene and microRNA module considering all three relations.
In addition, by applying an approach for identifying a gene-microRNA module having high cancer-relatedness according to the present invention to various cancer data, it can be useful for grasping genes, microRNAs, and biological pathways highly related to cancer.
BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram showing a step of collecting and pretreating gene and microRNA expression data. FIG.
FIG. 2 is a diagram illustrating a step of constructing a gene-cancer patient module utilizing bi-clustering.
Figure 3 is a diagram of the step of extending a gene to a gene-cancer patient module.
Fig. 4 is a diagram showing a step of constructing a gene-microRNA module. Fig.
5 is a diagram illustrating a step of verifying the association with cancer.
The present invention relates to a method for constructing a module representing the relationship between gene and gene expression data by integrating gene and microRNA expression data together with gene-gene interaction data.
First, we used a bi-clustering algorithm to construct a module that includes a subset of samples and a subset of genes to account for the heterogeneity of cancer cells.
Next, gene-gene interactions involving genes that play an important role in cancer-related biological pathways are combined.
Next, microRNAs closely related to the genes of the module were selected according to the Gaussian Bayesian network and the Bayesian Information Criteria.
As a result, the ovarian cancer data set consisted of 33 modules. In addition, the neuroblastoma (GBM) data set consisted of 54 modules.
In the 33 modules, 91% of the ovarian cancer modules were directly regulated between genes and microRNAs or indirectly mediated through transcription factors. In addition, in the 54 modules, 94% of the GBM modules were directly regulated between genes and microRNAs or indirectly regulated through transcription factors.
An extensive analysis of important modules revealed that most of the genes in the module were associated with ovarian cancer and GBM.
The present invention relates to a method and a device for collecting and pre-processing gene and microRNA expression data, constituting a gene-cancer patient module utilizing bi-clustering, expanding a gene to a gene-cancer patient module, And to an approach for identifying cancer-associated micro-RNA modules with cancer-associated stages of testing for their association with cancer.
Hereinafter, the present invention will be described in detail by way of examples.
The first step of collecting and pretreating gene and microRNA expression data will be described.
1) Ovarian cancer material
From the Cancer Genome Atlas (TCGA), mRNA expression and miRNA expression data sets for ovarian cancer, consisting of 587 tumor samples and 8 unmatched normal samples, were collected.
mRNA and miRNA expression data were generated using the Affymetrix HG-U133A microarray and the Agilent H-miRNA-8X15K microarray, respectively.
Referring to FIG. 1, for 12,042 genes, the expression level of each tumor sample was normalized using the
Likewise, for 479 miRNAs, the expression level of each tumor sample was normalized using the log2 ratio as an average of expression amounts of normal samples.
2) Nerve Gypsum material
MRNA expression and miRNA expression data sets for GBM, consisting of 482 tumor samples and 10 unmatched normal samples, were collected.
These data sets were generated using the same microarray platform used for ovarian cancer studies.
After normalization, 4,059 differentially expressed genes were selected using the t-test (the ferronically corrected p -value <0.05).
For 423 miRNAs, the expression levels of tumor samples were normalized using normal samples.
3) Select p-value threshold for t-test
The degree of expression varies with the type of cancer.
In the example, the number of differential expression genes was smaller in ovarian cancer than in GBM. Thus, we used a less stringent threshold for ovarian cancer.
4) gene-gene interactions
Gene-gene interaction data were collected from the HPRD database.
Next, the second step of constructing a gene-cancer patient module utilizing bi-clustering will be described.
In an embodiment, first, a gene group has a similar expression tendency to a subset of samples and hypothesizes that if the genes are differentially expressed, they may be related to similar functions or pathways in the development of cancer . Second, we hypothesized that genes have multiple functions and can function in multiple biological pathways.
To incorporate these hypotheses, the embodiment used a bi-clustering algorithm that allows duplication of genes and samples in multiple clusters.
First, the expression of each gene was normalized using a z-score to construct a matrix of differential gene expression and samples (matrix), and then determine the trend for changes in gene expression in the sample.
Next, referring to FIG. 2, the embodiment applies a SAMBA bi-clustering algorithm to a normalized matrix (matrix) to construct a module in which genes and samples are highly correlated.
In addition, the examples show that when genes and microRNAs are directly related (when the correlation coefficient has a negative value) and indirectly related by other factors such as transcription factor (when the correlation coefficient is positive And a module that can grasp all of them.
In this segmental graph G = (U, V, E) , the samba-bi-clustering algorithm model gene expression data is represented by the gene V on one side and the sample U on the opposite side. When the expression value of the gene v changes from sample u to a particularly high absolute value, there is an edge between the gene v at V and the sample u at U.
The bi-clustering algorithm generated a subgraph from this split graph. Most of the genes are connected to most of the samples as edges. The subgraph represents a highly correlated gene-sample cluster, and for a subset of the samples included in this module, the trends in the expression changes of genes are similar.
The amount of gene expression for the samples in the module is independent of the yield of other genes. The statistical significance of each module was calculated based on the null hypothesis and the mean Pearson correlation coefficients (PCCs) of gene expression levels in the module It should have a value greater than that obtained from a random module for the selected sample.
For each module, we performed the following tests.
Step 1 : Randomly select the same number of genes and samples from the normalized matrix to construct any module.
Step 2 : Compute the PCC matrix (matrix) of gene expression values in the module in a subset of samples. Then, except for the diagonal elements, the average value of the PCC matrix is calculated.
Step 3 :
Step 4 : The average PCC value of the gene in the observed module is module avg (i) .
Step 5 : Calculate the p value of the observed module using the following equation (1). Here, I is an indicator function.
When calculating the p-value, we tried to calculate the ratio of overlapping genes between observed modules that were not independent of each other. Thus, the present inventors have constructed a random module that shares the same redundancy ratio as the module in which the gene is observed in the module.
In other words, it is a method to construct a module by extracting genes showing similar gene expression patterns in some cancer patients from genetic expression information of cancer patients. In detail, this means the process of extracting a subset of gene-cancer patients from the entire gene-cancer patient data by inputting gene expression information into bi-clustering.
In the case of the conventional method, clustering can be performed only in one direction such as a set of genes or a set of cancer patients, but the present invention has an advantage that clustering is performed in a bi-directional module consisting of two elements of gene and cancer patient (sample) .
In addition, it is a method to construct such that genes and cancer patients (samples) can be duplicated in several modules. In detail, a gene appears in multiple modules, reflecting the biological principle that one gene is involved in multiple functions and pathways.
The embodiment allows the duplication of genes and samples in the module using the superposition coefficient of [0, 1] to 0.5, and applies the samba-bi-clustering algorithm to the gene expression matrix (matrix) to construct the gene-sample module. Where 1 represents non-overlapping. For ovarian cancer and GBM, changes in gene expression for a subset of samples, respectively, in the identified 90 and 135 modules, show similar trends.
Next, we describe the third step of gene extension to the gene-cancer patient module .
Recent studies have shown that not all of the genes in the cancer-related biological pathways undergo expression or genomic alterations. Thus, certain genes that play an important role in cancer-related biological pathways may not be differentially expressed.
In the gene-sample module, the gene-sample module was extended using a gene-gene interaction network to include functionally related genes.
When a gene interacts with one or more genes directly in a module, the gene can be considered a candidate gene for the module. For each module, the inventor collected candidate genes and calculated the average PCC value of expression between genes and candidate genes in the module.
Candidate genes were added to the module in descending order from genes with high PCC values. This task is repeated until the average PCC value of the expression of the genes contained in the module does not increase.
That is, referring to FIG. 3, the gene-sample (cancer patient) module includes gene interaction information. In detail, candidate genes that are highly related to each module were selected using gene interaction information. Taking into account the correlation coefficients, it is a method to include a gene with a high correlation with cancer in a gene-sample (cancer patient) module.
The example selected 58 and 88 modules with a q-value < 0.05 for ovarian cancer and GBM, respectively, after performing 1000 permutation tests. Next, the module is expanded by adding genes using gene-gene interactions. An average of 15 and 33 genes were added to each module for ovarian cancer and GBM.
Next, the fourth step of constructing the gene- miRNA module will be described.
Because genes with similar expression changes can be regulated by the same miRNA, gene-miRNA modules are constructed by including miRNAs that are regulated in gene sample modules. For this, Bayesian network model is used. The Bayesian network has been extensively used to analyze gene expression patterns. They are useful for modeling local dependency and causal influences between variables. Thus, the example estimated the dependency between miRNA expression values and gene expression values based on the Bayesian network model.
The joint distribution of the genes X = {X 1 , X 2 , ..., X n } and miRNAs Y = {Y 1 , Y 2 , ..., Y m } is expressed as a Gaussian Bayesian network do.
If X i is regularly distributed around its linearly dependent mean values in its parents, X i The conditional probability of X i The value of the parent of
The likelihood of X and Y can be expressed by the following equation (3).
In the gene-sample module, a Bayesian information criterion (BIC) was used as a measure to determine the Bayesian network structure between genes and miRNAs to determine the set of miRNAs that account for gene expression values, Can be expressed by Equation (4).
Where M is the sum of the number of genes and miRNAs. Parent of gene X i
However, this method is very time consuming.
To reduce the search area, candidate miRNAs are selected as an average of the absolute value of the Spearman's rank correlation coefficient (SCC) of genes in a given module in the upper T% of all miRNAs.
SCC values are used to select candidate miRNAs to reduce the effect of outliers on the PCC.
From the candidate miRNA, add the miRNA of the highest SCC value as a regulator and calculate the BIC score. Next, we add miRNAs with the next highest SCC value. This task is repeated until the BIC score is no longer improved.
After adding the miRNA to the gene-sample module using the above method, the module with fewer than two miRNAs is filtered because it can not exhibit gene and miRNA combination effects.
Finally, a gene miRNA module is obtained.
That is, referring to FIG. 4, a gene-microRNA module is constructed by constructing a Baysian network based on a gene-sample (cancer patient) module and calculating BIC. In detail, genes and microRNAs are modeled as Bayesian networks. By using the BIC score, it is a method of constructing a flexible module.
In an embodiment, the number of modules for ovarian cancer decreased with decreasing threshold. Similar results were obtained with GBM. Among the various threshold values of the candidate miRNA, the values of 3% (SCC value = 0.157 for ovarian cancer and 0.194 for GBM) were selected to constitute the gene-miRNA module and the 33 and 54 modules for ovarian cancer and GBM were selected respectively Respectively.
On average, in the ovarian cancer module, 19.1% of the genes were cancer genes, 5.7% were ovarian cancer genes, and 51.7% of miRNAs were ovarian cancer miRNAs. In the GBM module, 23.2% of the genes were cancer genes, 1.2% were GBM genes, and 71.7% of the miRNAs were GBM miRNA.
Since the example shows multiple genes and miRNAs appearing redundantly in several modules, the overlap ratio of genes and miRNAs between the modules was calculated.
The mean duplication rates of genes were 1.6% and 2.0% for ovarian cancer and GBM, respectively, and the average duplication rates for miRNA were 7.3% and 14.2% for ovarian cancer and GBM, respectively.
Next, the fifth step of verifying the association with cancer will be described.
To identify the relationship between genes and miRNAs in the module, four cases of gene regulation were considered.
In the first case, genes are directly regulated by miRNAs.
In the second case, the module considers the relationship that the miRNA regulates the transcription factor, and in the module the transcription factor regulates the gene. Here, the transfer factor need not be a member of the module.
In the third case, genes and miRNAs are regulated through common transcription factors. In this case, the correlation of the expression values between the gene-transcription factor and the miRNA-transcription factor is a positive correlation or a negative correlation.
In the fourth case, the interacting pair on miRTarbase was experimentally verified by co-expression of miRNA and mRNA and used to verify the gene-miRNA pair in the module.
Within the identified modules according to the examples, the ratio of experimentally proven interactions between genes and miRNAs was measured, as well as the direct relationship between genes and miRNAs and their indirect relationships through their transcription factors.
91% of the ovarian cancer modules (30 out of 33) and 94% (51 out of 54) of the GBM modules can be described for direct control or indirect relationships.
The genes involved in the development of cancer vary slightly depending on the type of cancer subtype. In several articles, the expression level of the target gene is used to determine the subtype.
For example, the GBM samples were classified as proneural subtypes when the marker genes DLL3, NKX2-2, SOX2, ERBB3, and OLIG2 were overexpressed.
Likewise, we identified the modules by our approach with specific subtypes of cancers using marker genes.
For this purpose, the following two steps were performed.
The first step is to cluster all samples into subtypes using hierarchical clustering with dynamic tree cuts. At this time, if known marker genes of the cancer subtype are overexpressed or underexpressed, each cluster is assigned to a subtype of cancer. If the cluster is not associated with any subtype or is associated with more than one subtype, then the cluster is not assigned to any subtype.
The second step is to calculate the significance using the t-test for each module, the difference between the amount of the sample contained in the other subtype cluster of each subtype marker gene and the average of the expression amount of the sample contained in the module Respectively. If the p-values of the marker genes of the subtype are important, consider that the module is associated with a given subtype.
That is, referring to FIG. 5, a general cancer gene, a gene related to a specific cancer, and a cancer gene associated with a specific cancer are identified through various databases. The association of the module with the cancer is statistically understood and correlated with the cancer. After constructing the gene-microRNA module, the association with cancer can be grasped through network analysis.
In the 33 modules according to the embodiment, 91% of the ovarian cancer modules were directly regulated between genes and microRNAs or indirectly regulated through transcription factors. In addition, in the 54 modules, 94% of the GBM modules were directly regulated between genes and microRNAs or indirectly regulated through transcription factors.
In addition, 48.4% and 74.0% of the modules from ovarian cancer and GBM, respectively, gave many stimuli in cancer-related biological pathways, and 51.7% and 71.7% of microRNAs in the module were ovarian cancer-related miRNA and GBM-related miRNA respectively .
We have analyzed the important modules extensively and found that most of the genes and miRNAs in these modules are associated with ovarian cancer and GBM.
The present invention can cluster similar genes based on the similarity of the genes appearing in some samples by clustering the gene-cancer patient (sample) module in two directions using samba bi-clustering. It is also possible to consider all cases where there is a negative correlation with the transfer factor and a positive correlation. Thus, the mechanism of various cancers can be grasped.
In addition, by applying an approach for identifying a gene-microRNA module having high cancer-relatedness according to the present invention to various cancer data, it can be useful for grasping genes, microRNAs, and biological pathways highly related to cancer.
While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. You will know that it is possible. For example, each component specifically shown in the embodiments can be modified and implemented. Further, it is to be understood that the scope of the present invention is defined by the appended claims. Therefore, the technical scope of the present invention should not be limited to the contents described in the detailed description of the specification, but should be defined by the claims.
Claims (7)
Constructing a gene-cancer sample module from a gene and microRNA expression data information by extracting a subset of whole gene expression data of a cancer patient using a bi-clustering algorithm;
Expanding the gene-cancer sample module by adding cancer-associated gene data to the gene-cancer sample module using a gene-gene interaction network; And
Constructing a gene-microRNA module by adding microRNA data to regulate a gene associated with the gene of the gene-cancer sample module using a Bayesian network calculated in the order of Equations 1 to 3; A method for processing data in a computer to identify a gene-microRNA module that is highly correlated with a cancer that performs a cancer.
<Formula 1>
P is a probability distribution,
N is a Gaussian distribution,
The joint distribution of genes X = {X 1 , X 2 , ..., X n } and miRNAs Y = {Y 1 , Y 2 , ..., Y m } is represented by a Bayesian network ,
If X i is normally distributed around a mean value, linearly dependent on its parent (parents), the conditional probability of X i is the value of the parents of X i
<Formula 2>
The likelihood (L, likelihood) of genes X and miRNAs Y is calculated by Equation 2,
<Formula 3>
In Equation 3, M is the sum of the numbers of genes and miRNAs,
O (1) is a constant term,
Parent of gene X i
The bi-clustering algorithm allows one gene to appear in multiple modules,
Wherein the gene-cancer sample module comprises a subset of normal genes and a subset of cancer samples, wherein the gene-microRNA module is associated with cancer.
Wherein expanding the gene-arm sample module comprises:
Calculating an average Pearson correlation coefficient between the gene data of the gene-arm sample module and the candidate gene data directly interacting with one or more genes in the gene-cancer sample module to determine, from the gene data having a high Pearson correlation coefficient value, A computer-implemented data processing method for identifying a gene-microRNA module that is highly related to cancer, wherein the candidate gene is added to the module until the mean Pearson correlation coefficient value of the gene expression included in the module does not increase .
A method of selecting genes and microRNAs that are highly related to cancer, according to computer data processing methods to identify gene-microRNA modules that are highly related to cancer.
A method of selecting genes and microRNAs highly correlated with cancer, which are highly correlated with ovarian cancer or glioma subtype, as microRNAs directly regulate gene expression or indirectly regulate gene expression.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020150110861A KR101816646B1 (en) | 2015-08-06 | 2015-08-06 | A METHOD FOR PROCESSING DATA OF A COMPUTER FOR IDENTIFYING GENE-microRNA MODULE HAVING HIGH COREELATION WITH CANCER AND A METHOD OF SELECTING GENES AND microRNAs HAVING HIGH COREELATION WITH CANCER |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020150110861A KR101816646B1 (en) | 2015-08-06 | 2015-08-06 | A METHOD FOR PROCESSING DATA OF A COMPUTER FOR IDENTIFYING GENE-microRNA MODULE HAVING HIGH COREELATION WITH CANCER AND A METHOD OF SELECTING GENES AND microRNAs HAVING HIGH COREELATION WITH CANCER |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| KR20170017231A KR20170017231A (en) | 2017-02-15 |
| KR101816646B1 true KR101816646B1 (en) | 2018-01-09 |
Family
ID=58112179
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| KR1020150110861A KR101816646B1 (en) | 2015-08-06 | 2015-08-06 | A METHOD FOR PROCESSING DATA OF A COMPUTER FOR IDENTIFYING GENE-microRNA MODULE HAVING HIGH COREELATION WITH CANCER AND A METHOD OF SELECTING GENES AND microRNAs HAVING HIGH COREELATION WITH CANCER |
Country Status (1)
| Country | Link |
|---|---|
| KR (1) | KR101816646B1 (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101959060B1 (en) * | 2017-04-19 | 2019-03-18 | 광주과학기술원 | A module constructing method for studying cancer |
| KR102188118B1 (en) * | 2019-04-15 | 2020-12-07 | 인천대학교 산학협력단 | Electronic device for generating a gene feature vector for gene distributed representation based on a correlation between genes according to cancer and operating method thereof |
-
2015
- 2015-08-06 KR KR1020150110861A patent/KR101816646B1/en active IP Right Grant
Non-Patent Citations (1)
| Title |
|---|
| Bioinformatics, Vol.26, No.18, pages i638-i644, (2010.9월). |
Also Published As
| Publication number | Publication date |
|---|---|
| KR20170017231A (en) | 2017-02-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| AU2022268283B2 (en) | Phenotype/disease specific gene ranking using curated, gene library and network based data structures | |
| Assefa et al. | Differential gene expression analysis tools exhibit substandard performance for long non-coding RNA-sequencing data | |
| Li et al. | Putative biomarkers for predicting tumor sample purity based on gene expression data | |
| Zeng et al. | Cancer classification and pathway discovery using non-negative matrix factorization | |
| Guo et al. | A novel network model identified a 13-gene lung cancer prognostic signature | |
| Bandyopadhyay et al. | A biologically inspired measure for coexpression analysis | |
| Park et al. | Sparse overlapping group lasso for integrative multi-omics analysis | |
| Ly et al. | Effect of imputation on gene network reconstruction from single-cell RNA-seq data | |
| Yu et al. | Comparing five statistical methods of differential methylation identification using bisulfite sequencing data | |
| Tsirvouli et al. | A middle-out modeling strategy to extend a colon cancer logical model improves drug synergy predictions in epithelial-derived cancer cell lines | |
| KR101816646B1 (en) | A METHOD FOR PROCESSING DATA OF A COMPUTER FOR IDENTIFYING GENE-microRNA MODULE HAVING HIGH COREELATION WITH CANCER AND A METHOD OF SELECTING GENES AND microRNAs HAVING HIGH COREELATION WITH CANCER | |
| KR20210110241A (en) | Prediction system and method of cancer immunotherapy drug Sensitivity using multiclass classification A.I based on HLA Haplotype | |
| Lock et al. | Bayesian genome-and epigenome-wide association studies with gene level dependence | |
| Yang et al. | Improving clustering of microRNA microarray data by incorporating functional similarity | |
| Guzzi et al. | A discussion on the biological relevance of clustering results | |
| Yang et al. | MSPL: Multimodal self-paced learning for multi-omics feature selection and data integration | |
| Rudra et al. | Multiple testing approaches for hypotheses in integrative genomics | |
| Zhao et al. | Ensemble classification based signature discovery for cancer diagnosis in RNA expression profiles across different platforms | |
| De Franchis | Design of a CNN-based method for classifying subtype of kidney cancers using miRNA isoform profiles | |
| Mai et al. | Identification of biomarkers for predicting the overall survival of ovarian cancer patients: a sparse group lasso approach | |
| Arango et al. | AI-based predictive biomarker discovery via contrastive learning retrospectively improves clinical trial outcome | |
| VIEIRA | Unveiling Novel Glioma Biomarkers through Multi-omics Integration and Classification | |
| Zhang | Computational Analysis of Transcript Interactions and Variants in Cancer | |
| Kacar | Dissecting Tumor Clonality in Liver Cancer: A Phylogeny Analysis Using Computational and Statistical Tools | |
| Ye | A Novel Computational Network Methodology for Discovery of Biomarkers and Therapeutic Targets |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| A201 | Request for examination | ||
| E902 | Notification of reason for refusal | ||
| GRNT | Written decision to grant |