A Novel XGBoost Method to Infer the Primary Lesion of 20 Solid Tumor Types From Gene Expression Data.
Ontology highlight
ABSTRACT: Purpose
Establish a suitable machine learning model to identify its primary lesions for primary metastatic tumors in an integrated learning approach, making it more accurate to improve primary lesions' diagnostic efficiency.Methods
After deleting the features whose expression level is lower than the threshold, we use two methods to perform feature selection and use XGBoost for classification. After the optimal model is selected through 10-fold cross-validation, it is verified on an independent test set.Results
Selecting features with around 800 genes for training, the R 2-score of a 10-fold CV of training data can reach 96.38%, and the R 2-score of test data can reach 83.3%.Conclusion
These findings suggest that by combining tumor data with machine learning methods, each cancer has its corresponding classification accuracy, which can be used to predict primary metastatic tumors' location. The machine-learning-based method can be used as an orthogonal diagnostic method to judge the machine learning model processing and clinical actual pathological conditions.
SUBMITTER: Chen S
PROVIDER: S-EPMC7886791 | biostudies-literature | 2021
REPOSITORIES: biostudies-literature
Publications
A Novel XGBoost Method to Infer the Primary Lesion of 20 Solid Tumor Types From Gene Expression Data.
Chen Sijie S Zhou Wenjing W Tu Jinghui J Li Jian J Wang Bo B Mo Xiaofei X Tian Geng G Lv Kebo K Huang Zhijian Z
Frontiers in genetics 20210203
<h4>Purpose</h4>Establish a suitable machine learning model to identify its primary lesions for primary metastatic tumors in an integrated learning approach, making it more accurate to improve primary lesions' diagnostic efficiency.<h4>Methods</h4>After deleting the features whose expression level is lower than the threshold, we use two methods to perform feature selection and use XGBoost for classification. After the optimal model is selected through 10-fold cross-validation, it is verified on ...[more]
Similar Datasets
Project description:BackgroundGlucocorticoid receptor (GR) activity plays a role in many aspects of human physiology and may play a crucial role in chemotherapy resistance in a wide variety of solid tumors. A novel immunohistochemistry (IHC) based assay has been previously developed and validated in order to assess GR immunoreactivity in triple-negative breast cancer. The current study investigates the standardized use of this validated assay to assess GR expression in a broad range of solid tumor malignancies.MethodsArchived formalin-fixed paraffin-embedded tumor bank samples (n=236) from 20 different solid tumor types were analyzed immunohistochemically. Nuclear staining was reported based on the H-score method using differential intensity scores (0, 1+, 2+, or 3+) with the percent stained (out of at least 100 carcinoma cells) recorded at each intensity.ResultsGR was expressed in all tumor types that had been evaluated. Renal cell carcinoma, sarcoma, cervical cancer, and melanoma were those with the highest mean H-scores, indicating high levels of GR expression. Colon, endometrial, and gastric cancers had lower GR staining percentages and intensities, resulting in the lowest mean H-scores.ConclusionA validated IHC assay revealed GR immunoreactivity in all solid tumor types studied and allowed for standardized comparison of reactivity among the different malignancies.ImpactBaseline expression levels of GR may be a useful biomarker when pharmaceutically targeting GR in research or clinical setting.
| S-EPMC5345989 | biostudies-literature
Project description:Cancer of unknown primary site (CUP) is a heterogeneous group of cancers whose tissue of origin remains unknown after detailed investigation by conventional clinical methods. The number of CUP accounts for roughly 3%-5% of all human malignancies. CUP patients are usually treated with broad-spectrum chemotherapy, which often leads to a poor prognosis. Recent studies suggest that the treatment targeting the primary lesion of CUP will significantly improve the prognosis of the patient. Therefore, it is urgent to develop an efficient method to accurately detect tissue of origin of CUP in clinical cancer research. In this work, we developed a novel framework that uses Extreme Gradient Boosting (XGBoost) to trace the primary site of CUP based on microarray-based gene expression data. First, we downloaded the microarray-based gene expression profiles of 59,385 genes for 57,08 samples from The Cancer Genome Atlas (TCGA) and 6,364 genes for 3,101 samples from the Gene Expression Omnibus (GEO). Both data were divided into training and independent testing data with a ratio of 4:1. Then, we obtained in the training data 200 and 290 genes from TCGA and the GEO datasets, respectively, to train XGBoost models for the identification of the primary site of CUP. The overall 5-fold cross-validation accuracies of our methods were 96.9% and 95.3% on TCGA and GEO training datasets, respectively. Meanwhile, the macro-precision for the independent dataset reached 96.75% and 98.8% on, respectively, TCGA and GEO. Experimental results demonstrated that the XGBoost framework not only can reduce the cost of clinical cancer traceability but also has high efficiency, which might be useful in clinical usage.
| S-EPMC9071249 | biostudies-literature
Project description:BackgroundMachine learning (ML) represents a powerful tool to capture relationships between molecular alterations and cancer types and to extract biological information. Here, we developed a plain ML model aimed at distinguishing cancer types based on genetic lesions, providing an additional tool to improve cancer diagnosis, particularly for tumors of unknown origin.MethodsTCGA data from 9,927 samples spanning 32 different cancer types were downloaded from cBioportal. A vector space model type data transformation technique was designed to build consistently homogeneous new datasets containing, as predictive features, calls for somatic point mutations and copy number variations at chromosome arm-level, thus allowing the use of the XGBoost classifier models. Considering the imbalance in the dataset, due to large difference in the number of cases for each tumor, two preprocessing strategies were considered: i) setting a percentage cut-off threshold to remove less represented cancer types, ii) dividing cancer types into different groups based on biological criteria and training a specific XGBoost model for each of them. The performance of all trained models was mainly assessed by the out-of-sample balanced accuracy (BACC) and the AUC scores.ResultsThe XGBoost classifier achieved the best performance (BACC 77%; AUC 97%) on a dataset containing the 10 most represented tumor types. Moreover, dividing the 18 most represented cancers into three different groups (endocrine-related carcinomas, other carcinomas and other cancers),such analysis models achieved 78%, 71% and 86% BACC, respectively, with AUC scores greater than 96%. In addition, the model capable of linking each group to a specific cancer type reached 81% BACC and 94% AUC. Overall, the diagnostic potential of our model was comparable/higher with respect to others already described in literature and based on similar molecular data and ML approaches.ConclusionsA boosted ML approach able to accurately discriminate different cancer types was developed. The methodology builds datasets simpler and more interpretable than the original data, while keeping enough information to accurately train standard ML models without resorting to sophisticated Deep Learning architectures. In combination with histopathological examinations, this approach could improve cancer diagnosis by using specific DNA alterations, processed by a replicable and easy-to-use automated technology. The study encourages new investigations which could further increase the classifier's performance, for example by considering more features and dividing tumors into their main molecular subtypes.
| S-EPMC10664515 | biostudies-literature
Project description:BackgroundOsteomimicry of cancer cells had been widely reported in prostate cancer and breast cancer. However, the prognostic value of osteomimicry in various cancer types remained unclear. We hypothesized that osteomimicry would result in remodeling of the tumor microenvironment and was eligible to predict patient prognosis.MethodsA comprehensive transcriptomic analysis of the osteomimicry, which was characterized by mRNA expression of SPARC, SPP1, and BGLAP, across 20 solid tumors (7564 patients) using RNA-seq data from The Cancer Genome Atlas (TCGA) was conducted. Samples of each cancer type were classified into subgroups (high vs. low) based on median value of osteomimetic markers, the associations of these markers with clinical outcomes, immune cell infiltration and immune checkpoints expression were explored.ResultsEach osteomimetic marker harbored prognostic value in the pan-cancer analyses [SPARC: hazard ratio (HR) = 1.10, p = 0.028; SPP1: HR = 1.25, p < 0.001; BGLAP: HR = 1.13, p = 0.005]. Patients with high expression of all the three genes also had significantly unfavorable survival (HR = 1.61, p < 0.0001) compared with those of low expression. Correlation analyses demonstrated that osteomimicry was closely related to tumor purity, dendritic cells (DC) infiltration and expression of immune checkpoints.ConclusionOsteomimicry had prognostic value in various cancer types and the underlying mechanism might correlate to the trapping and dysfunction of DCs in the tumor microenvironment, revealing the potential of osteomimicry as a target of immunotherapy.
| S-EPMC7678014 | biostudies-literature
Project description:The identification of good targets is a critical step for the development of targeted therapies for cancer treatment. Here, we used a multi-omics approach to delineate potential targets on chromosome 20q, which frequently shows a complex pattern of DNA copy number amplification in many human cancers suggesting the presence of multiple driver genes. By comparing the amounts of individual mRNAs in cancer from 11 different human tissues with those in their corresponding normal tissues, we identified 18 genes that were robustly elevated across human cancers. Moreover, we found that higher expression levels of a majority of these genes were associated with poor prognosis in many human cancer types. Using DNA copy number and expression data for all 18 genes obtained from The Cancer Genome Atlas project, we discovered that amplification is a major mechanism driving overexpression of these 18 genes in the majority of human cancers. Our integrated analysis suggests that 18 genes on chromosome 20q might serve as novel potential molecular targets for targeted cancer therapy.
| S-EPMC5025263 | biostudies-literature
Project description:Reconstructing gene regulatory networks is crucial to understand biological processes and holds potential for developing personalized treatment. Yet, it is still an open problem as state-of-the-art algorithms are often not able to process large amounts of data within reasonable time. Furthermore, many of the existing methods predict numerous false positives and have limited capabilities to integrate other sources of information, such as previously known interactions. Here we introduce KBoost, an algorithm that uses kernel PCA regression, boosting and Bayesian model averaging for fast and accurate reconstruction of gene regulatory networks. We have benchmarked KBoost against other high performing algorithms using three different datasets. The results show that our method compares favorably to other methods across datasets. We have also applied KBoost to a large cohort of close to 2000 breast cancer patients and 24,000 genes in less than 2 h on standard hardware. Our results show that molecularly defined breast cancer subtypes also feature differences in their GRNs. An implementation of KBoost in the form of an R package is available at: https://github.com/Luisiglm/KBoost and as a Bioconductor software package.
| S-EPMC8322418 | biostudies-literature
Project description:We performed a comprehensive immuno-genomic analysis of tumor microenvironment immune types (TMITs), which is classified into four groups based on PD-L1+CD8A or PD-L1+cytolytic activity (CYT) expression, across a broad spectrum of solid tumors in order to help identify patients who will benefit from anti- PD-1/PD-L1 therapy. The mRNA sequencing data from The Cancer Genome Atlas (TCGA) of 14 solid cancer types representing 6,685 tumor samples was analyzed. TMIT was classified only for those tumor types that both PD-L1 and CD8A/CYT could prefict mutation and/or neoantigen number. The mutational and neoepitope features of the tumor were compared according to the four TMITs. We found that PD-L1/CD8A/CYT subgroups could not distinguish different mutation and neoantigen numbers in certain tumor types such as glioblastoma multiforme, prostate adenocarcinoma, and head and neck and lung squamous cell carcinoma. For the remaining tumor types, compared with TIMT II (low PD-L1 and CD8A/CYT), TIMT I (high PD-L1 and CD8A/CYT) had a significantly higher number of mutations or neoantigens in bladder urothelial carcinoma, breast and cervical cancer, colorectal, stomach and lung adenocarcinoma, and melanoma. In contrast, TMIT I of kidney clear cell, liver hepatocellular, and thyroid carcinoma were negatively correlated with mutation burden or neoantigen numbers. Our findings show that the TMIT stratification proposed could serve as a favorable approach for tailoring optimal immunotherapeutic strategies in certain tumor types. Going forward, it will be important to test the clinical practicability of TMIT based on quantification of immune infiltrates using mRNA-seq to predict clinical response to these and other immunotherapeutic strategies in more different tumors.
| S-EPMC5596445 | biostudies-literature
Project description:Knowledge of miRNA expression and function in specific cell types in solid organs is limited because of difficulty in obtaining appropriate specimens. We used laser capture microdissection to obtain nine tissue regions from rats, including the nucleus of the solitary tract, hypoglossal motor nucleus, ventral respiratory column/pre-Bötzinger complex, and midline raphe nucleus from the brain stem, myocardium and coronary artery from the heart, and glomerulus, proximal convoluted tubule, and medullary thick ascending limb from the kidney. Each tissue region consists of or is enriched for a specific cell type. Differential patterns of miRNA expression obtained by deep sequencing of minute amounts of laser-captured cells were highly consistent with data obtained from real-time PCR analysis. miRNA expression patterns correctly clustered the specimens by tissue regions and then by primary tissue types (neural, muscular, or epithelial). The aggregate difference in miRNA profiles between tissue regions that contained the same primary tissue type was as large as one-half of the aggregate difference between primary tissue types. miRNAs differentially expressed between primary tissue types are more likely to be abundant miRNAs, while miRNAs differentially expressed between tissue regions containing the same primary tissue type were distributed evenly across the abundance spectrum. The tissue type-enriched miRNAs were more likely to target genes enriched for specific functional categories compared with either cell type-enriched miRNAs or randomly selected miRNAs. These data indicate that the role of miRNAs in determining characteristics of primary tissue types may be different than their role in regulating cell type-specific functions in solid organs.
| S-EPMC3882707 | biostudies-literature
Project description:Integrins are contributors to remodeling of the extracellular matrix and cell migration. Integrins participate in the assembly of the actin cytoskeleton, regulate growth factor signaling pathways, cell proliferation, and control cell motility. In solid tumors, integrins are involved in promoting metastasis to distant sites, and angiogenesis. Integrins are a key target in cancer therapy and imaging. Integrin antagonists have proven successful in halting invasion and migration of tumors. Overexpressed integrins are prime anti-cancer drug targets. To streamline the development of specific integrin cancer therapeutics, we curated data to predict which integrin heterodimers are pausible therapeutic targets against 17 different solid tumors. Computational analysis of The Cancer Genome Atlas (TCGA) gene expression data revealed a set of integrin targets that are differentially expressed in tumors. Filtered by FPKM (Fragments Per Kilobase of transcript per Million mapped reads) expression level, overexpressed subunits were paired into heterodimeric protein targets. By comparing the RNA-seq differential expression results with immunohistochemistry (IHC) data, overexpressed integrin subunits were validated. Biologics and small molecule drug compounds against these identified overexpressed subunits and heterodimeric receptors are potential therapeutics against these cancers. In addition, high-affinity and high-specificity ligands against these integrins can serve as efficient vehicles for delivery of cancer drugs, nanotherapeutics, or imaging probes against cancer.
| S-EPMC6059022 | biostudies-literature
Project description:Identifying subpopulations within a study and inferring intercontinental ancestry of the samples are important steps in genome wide association studies. Two software packages are widely used in analysis of substructure: Structure and Eigenstrat. Structure assigns each individual to a population by using a Bayesian method with multiple tuning parameters. It requires considerable computational time when dealing with thousands of samples and lacks the ability to create scores that could be used as covariates. Eigenstrat uses a principal component analysis method to model all sources of sampling variation. However, it does not readily provide information directly relevant to ancestral origin; the eigenvectors generated by Eigenstrat are sample specific and thus cannot be generalized to other individuals.We developed FastPop, an efficient R package that fills the gap between Structure and Eigenstrat. It can: 1, generate PCA scores that identify ancestral origins and can be used for multiple studies; 2, infer ancestry information for data arising from two or more intercontinental origins. We demonstrate the use of FastPop using 2318 SNP markers selected from the genome based on high variability among European, Asian and West African (African) populations. We conducted an analysis of 505 Hapmap samples with European, African or Asian ancestry along with 19661 additional samples of unknown ancestry. The results from FastPop are highly consistent with those obtained by Structure across the 19661 samples we studied. The correlations of the results between FastPop and Structure are 0.99, 0.97 and 0.99 for European, African and Asian ancestry scores, respectively. Compared with Structure, FastPop is more efficient as it finished ancestry inference for 19661 samples in 16 min compared with 21-24 h required by Structure. FastPop also provided scores based on SNP weights so the scores of reference population can be applied to other studies provided the same set of markers are used. We also present application of the method for studying four continental populations (European, Asian, African, and Native American).We developed an algorithm that can infer ancestries on data involving two or more intercontinental origins. It is efficient for analyzing large datasets. Additionally the PCA derived scores can be applied to multiple data sets to ensure the same ancestry analysis is applied to all studies.
| S-EPMC4784403 | biostudies-literature