Racial Disparity in the Genomic Basis of Radiosensitivity – An Exploration of Whole-Transcriptome Sequencing Data via a Machine-Learning Approach

Abstract Fusion transcripts are frequent genetic abnormalities in myeloid malignancies and are often the basis for risk stratification, minimal residual disease (MRD) monitoring, and targeted therapy. We comprehensively analyzed the fusion transcript landscape in 572 acute myeloid leukemia (AML) and 630 myelodysplastic syndrome (MDS) patients by whole transcriptome sequencing (WTS). Totally, 274 fusion events (131 unique fusions) were identified in 210/572 AML patients (37%). In 16/630 MDS patients, 16 fusion events (15 unique fusions) were detected (3%). In AML, 141 cases comprised entity-defining rearrangements (51% of all detected fusions) and 21 (8%) additional well-known fusions, all detected by WTS (control group). In MDS, only 1 fusion was described previously (NRIP1-MECOM, n = 2). Interestingly, a high number of so-far unreported fusions were found (41% [112/274] in AML, 88% [14/16] in MDS), all validated by cytogenetic and/or whole genome sequencing data. With 1 exception (CTDSP1-CFLAR, n = 2), all novel fusions were observed in 1 patient each. In AML, cases with novel fusions showed concomitantly a high frequency of TP53 mutations (67%) and of a complex karyotype (71%), which was also observed in MDS, but less pronounced (TP53, 26%; complex karyotype, 21%). A functional annotation of genes involved in novel fusions revealed many functional relevant genes (eg, transcription factors; n = 28 in AML, n = 2 in MDS) or enzymes (n = 42 in AML, n = 9 in MDS). Taken together, new genomic alterations leading to fusion transcripts were much more common in AML than in MDS. Any novel fusions might be of use for developing markers (eg, for MRD monitoring), particularly in cases without an entity-defining abnormality.

Download Full-text

Finding small somatic structural variants in exome sequencing data: a machine learning approach

Computational Statistics ◽

10.1007/s00180-016-0674-2 ◽

2016 ◽

Vol 33 (3) ◽

pp. 1145-1158 ◽

Cited By ~ 2

Author(s):

Matthias Kuhn ◽

Thoralf Stange ◽

Sylvia Herold ◽

Christian Thiede ◽

Ingo Roeder

Keyword(s):

Machine Learning ◽

Exome Sequencing ◽

Learning Approach ◽

Structural Variants ◽

Sequencing Data ◽

Exome Sequencing Data ◽

Machine Learning Approach

Download Full-text

Integrated Analysis of lncRNA-associated ceRNA Network Identifies Prognostic lncRNAs in Bladder Cancer Based on Whole Transcriptome Sequencing and TCGA Database

10.21203/rs.3.rs-663356/v1 ◽

2021 ◽

Author(s):

Shiwei Xiao ◽

Yigang Zuo ◽

Yinglong Huang ◽

Yanan Li ◽

Dongbo Yuan ◽

...

Keyword(s):

Risk Factors ◽

Transcriptome Sequencing ◽

Risk Groups ◽

Gene Signature ◽

Stepwise Multiple Regression ◽

Sequencing Data ◽

Predicting Model ◽

Whole Transcriptome Sequencing ◽

Independent Risk Factors ◽

Whole Transcriptome

Abstract BackgroundBladder cancer (BC) is a serious urinary tract malignancy with high incidences and deaths. Here, we aim to explore the OS-related lncRNAs and mRNAs, and constructed a reliable predicting model for accurately predict BC prognosis. Methods3 matched BC tumor samples and adjacent normal samples were processed by whole transcriptome sequencing. Differentially expressed lncRNAs and mRNAs (DE-lncRNAs and DE-mRNAs) were identified with the thresholds |log2 (FC) |>2 and p value <0.05 using DEGseq2 R package. Meantime, 408 BC samples and 19 normal samples were downloaded from TCGA database. And DE-lncRNAs and DE-mRNAs were screened with thresholds |log2(FC) |>0.5 and p value<0.05. The DE-lncRNAs and DE-mRNAs were overlapped between RNA-sequencing data and TCGA data. Then, overall survival (OS)-related DE-lncRNAs and DE-mRNAs were screened and a prognostic gene signature and risk model were constructed using Univariate Cox and stepwise multiple regression analysis. Risk score was calculated based on prognostic gene signature. The independent risk factors were identified with incorporation into clinical risk factors using Univariate Cox and stepwise multiple regression analysis. A predicting model was constructed based on independent factors, and calibrated using Time-dependent ROC curves. Finally, the differentially expressed genes (DEGs) between high-risk and low-risk groups were identified with thresholds |log2(FC) |>0.5 and p value<0.05. And the biofunction was determined by Gene Set Enrichment Analysis (GSEA). ResultsA total 2210 DE-lncTNAs and 2334 DE-mRNAs were identified based on whole transcriptome sequencing. And a total 3724 DE-lncRNAs and 2689 DE-mRNAs were identified based on TCGA database. Then, 137 DE-lncRNAs and 278 DE-mRNAs were screened by overlapping RNA sequencing data and TCGA data. A total 13 gene signature were identified to be closely related with OS in BC. Moreover, these 13-OS-related gene signature were identified as independent risk factors with clinical risk factors. Then, a nomogram was constructed and confirmed as the calibration and accuracy predicating model for OS in BC. Finally, a total 739 DEGs were identified between high-risk and low-risk groups, most of DEGs enriched in immune-related pathways. And an OS-related ceRNA network was selected based on 13-OS-related signature. ConclusionOur finding provided novel OS-related prognostic signature and reliable predicting model for BC patients, which might facilitate individualized treatment and prognostic evaluation.

Download Full-text