Data for Non-Neotissue Constituents as Underestimated Confounders in the Assessment of Tissue Engineered Constructs by Near-Infrared Spectroscopy

Datasets, orange workflow file and R scripts necessary to replicate the results of the study "Non-Neotissue Constituents as Underestimated Confounders in the Assessment of Tissue Engineered Constructs by Near-Infrared Spectroscopy".
S1-Samples.csv: The details of the samples used in the study.
S2-Spectra.csv: The labeled spectra used for data analysis.
S3-OrangeWorkFlow.ows: The orange workflow file used for primary machine learning as mentioned in section 2.3 in the study. Requires the software Quasar or Orange 3.33 with the Quasar add-on (Bioinformatics Lab, University of Ljubljana, Slovenia). Upload the S2-Spectra.csv file in the "file" widget for the analysis to start (may require manual intiation in some widgets too).
S4.Orangeworkflows.pdf: Detailed overview of the S3-OrangeWorkFlow.ows.
S5-Propensity Score Matching.zip: Files used for Propensity Score Matching mentioned in section 2.3.2
S6-matchedmatrix.csv: The matched pair from Propensity Score Matching produced from "S5-Propensity Score Matching.zip" and used for controlling the models in "S3-OrangeWorkFlow.ows" and in "S7-MCCVofSVM.Rmd" as described in section 2.3.2.
S7-MCCVofSVM.Rmd: Used for Monte Carlo cross-validation mentioned in section 2.3.2 and produce figure 4
S8-ConfoundersPredictionPerformances.xlsx: The machine learning cross-validation results for models classifying the constructs according to the non-neotissue constituents. Table 3 of the study presents summary of selected metrics form these spreadsheet.
S9-ControlledVsUncontrolledPerformances.xlsx: The comprehensive results showing all metrics and the statistical analysis results of the controlled vs the uncontrolled models mentioned in section 2.3.2 and described in section 3.2
S10- Statistics: Files and results of the statistical analysis described in section 3.2 and presented in Table 4

本数据集包含复现研究《近红外光谱（Near-Infrared Spectroscopy）评估组织工程构建物（Tissue Engineered Constructs）时非新组织成分作为被低估混杂因素》所需的数据集、Orange工作流文件与R脚本。
S1-Samples.csv：本文件收录本研究使用的样本详细信息。
S2-Spectra.csv：本文件包含用于数据分析的标注光谱数据集。
S3-OrangeWorkFlow.ows：本文件为研究第2.3节所述的基础机器学习所用的Orange工作流文件，需搭配Quasar软件或搭载Quasar插件（斯洛文尼亚卢布尔雅那大学生物信息实验室开发）的Orange 3.33版本运行。需在"文件"组件中导入S2-Spectra.csv文件以启动分析（部分组件可能需手动触发）。
S4.Orangeworkflows.pdf：对S3-OrangeWorkFlow.ows的详细说明文档。
S5-Propensity Score Matching.zip：包含研究第2.3.2节所述的倾向得分匹配（Propensity Score Matching）所需文件。
S6-matchedmatrix.csv：本文件为从S5-Propensity Score Matching.zip中导出的倾向得分匹配匹配对数据集，用于第2.3.2节所述的S3-OrangeWorkFlow.ows与S7-MCCVofSVM.Rmd中的模型控制环节。
S7-MCCVofSVM.Rmd：用于复现第2.3.2节所述的蒙特卡洛交叉验证（Monte Carlo cross-validation），并生成研究图4。
S8-ConfoundersPredictionPerformances.xlsx：收录基于非新组织成分对组织工程构建物进行分类的机器学习交叉验证结果。研究表3汇总了该电子表格中部分选定指标的结果。
S9-ControlledVsUncontrolledPerformances.xlsx：收录全面的模型性能指标与统计分析结果，对应第2.3.2节所述、第3.2节展示的受控模型与非受控模型对比分析结果。
S10-Statistics：包含第3.2节所述、表4展示的统计分析所需文件与最终结果。