Analysis of Half a Billion Datapoints Across Ten Machine-Learning Algorithms Identifies Key Elements Associated With Insulin Transcription in Human Pancreatic Islet Cells.

Analysis of Half a Billion Datapoints Across Ten Machine-Learning Algorithms Identifies Key Elements Associated With Insulin Transcription in Human Pancreatic Islet Cells.
复制标题

DOI:
10.3389/fendo.2022.853863
复制
发表时间:
2022
影响因子:
5.2
通讯作者:
Hardikar AA
Hardikar AA
中科院分区:
医学2区
文献类型:
--
作者:
Wong WKM;Thorat V;Joglekar MV;Dong CX;Lee H;Chew YV;Bhave A;Hawthorne WJ;Engin F;Pant A;Dalgaard LT;Bapat S;Hardikar AA

文献摘要

参考文献

相似文献

机器学习 (ML) 工作流程可以对复杂数据集进行不带偏见/稳健的评估。在这里,我们分析了超过 490,000,000 个数据点,以比较人类胰腺单细胞 (sc-) 转录组的大型 (N=11,652) 训练数据集中的 10 个不同的 ML 工作流程,以确定与胰岛素转录物存在或不存在相关的基因。每个 ML 工作流程的预测准确性/灵敏度在单独的验证数据集中进行测试 (N=2,913)。与其他算法相比,Ensemble ML 工作流程,特别是随机森林 ML 算法,具有较高的预测能力 (AUC=0.83) 和灵敏度 (0.98)。通过这些分析鉴定的转录本还证明与人类胰岛的大量 RNA-seq 数据中的胰岛素存在显着相关性。三个 Ensemble ML 工作流程共有的前 10 个特征(包括 IAPP、ADCYAP1、LDHA 和 SST)在 Ire-1αβ-/- 小鼠的 scRNA-seq 数据集中显着失调,这些数据表明 1 型糖尿病 (T1D) 模型中的胰腺 β 细胞和 2 型糖尿病 (T2D) 个体的胰腺单细胞中的胰腺 β 细胞去分化。我们的研究结果提供了大数据分析中机器学习工作流程的直接比较,确定了与胰岛素转录相关的关键要素,并为未来的分析提供了工作流程。
Machine learning (ML)-workflows enable unprejudiced/robust evaluation of complex datasets. Here, we analyzed over 490,000,000 data points to compare 10 different ML-workflows in a large (N=11,652) training dataset of human pancreatic single-cell (sc-)transcriptomes to identify genes associated with the presence or absence of insulin transcript(s). Prediction accuracy/sensitivity of each ML-workflow was tested in a separate validation dataset (N=2,913). Ensemble ML-workflows, in particular Random Forest ML-algorithm delivered high predictive power (AUC=0.83) and sensitivity (0.98), compared to other algorithms. The transcripts identified through these analyses also demonstrated significant correlation with insulin in bulk RNA-seq data from human islets. The top-10 features, (including IAPP, ADCYAP1, LDHA and SST) common to the three Ensemble ML-workflows were significantly dysregulated in scRNA-seq datasets from Ire-1αβ-/- mice that demonstrate dedifferentiation of pancreatic β-cells in a model of type 1 diabetes (T1D) and in pancreatic single cells from individuals with type 2 Diabetes (T2D). Our findings provide direct comparison of ML-workflows in big data analyses, identify key elements associated with insulin transcription and provide workflows for future analyses.
DOI: 10.1002/humu.9034
发表时间: 2002-05-01
期刊: Human mutation
影响因子: 3.9
作者:
Gu, Harvest F
通讯作者: Gu, Harvest F
DOI: 10.1016/j.cmet.2020.03.002
发表时间: 2020-04-07
期刊: CELL METABOLISM
影响因子: 29
作者:
Lee, Hugo;Lee, Yong-Syu;Engin, Feyza
通讯作者: Engin, Feyza
DOI: 10.1371/journal.pmed.1003112
发表时间: 2020-05-01
期刊: PLOS MEDICINE
影响因子: 15.8
作者:
Lai, Mi;Liu, Ying;Wheeler, Michael B.
通讯作者: Wheeler, Michael B.
DOI: 10.1038/12674
发表时间: 1999-09-01
期刊: NATURE GENETICS
影响因子: 30.8
作者:
Harrison, KA;Thaler, J;Kehrl, JH
通讯作者: Kehrl, JH
DOI: 10.1161/atvbaha.118.311298
发表时间: 2018-09-01
影响因子: 8.7
作者:
Badi, Ileana;Mancinelli, Luigi;Raucci, Angela
通讯作者: Raucci, Angela