Original networks, NVC networks and COPD data sets used in: Enhancement of COPD biological networks using a web-based collaboration interface

Original networks, NVC networks and their descriptions.<br>The file contains the names of the original networks (as they were published), agglomerated NVC networks (as presented on the Bionet website), and network descriptions. The 15 networks that were discussed during jamboree are indicated by “X” in the column Discussed in Jamboree.
COPD data sets, their descriptions, and the comparisons used to build the COPD models during Phase 1.<br>Reverse causal reasoning was performed using COPD and emphysema data sets from lung, small airway, and alveolar macrophages of early COPD patients and healthy smokers. Data Sets, the Gene Expression Omnibus (GEO) used to build the COPD networks. SCs, state changes defined using differentially expressed genes that meet the following criteria: FDR adjusted p&lt;0.05, fold change ≥1.3, and minimum expression of 100 (for Affy platforms). HYPs, mechanisms or hypotheses predicted from the SCs and the Selventa Knowledgebase [1] with the following cutoffs: richness p&lt;0.1, concordance p&lt;0.1.
Early COPD was defined as Global Initiative for Chronic Obstructive Lung Disease (GOLD) stages 1 and 2.<br>The three small airway data sets were merged using ComBat [2] because of the small sample size of early COPD patients within each data set.<br>Lone emphysema is defined in the GSE10006 data set as patients who have normal spirometry but decreased transfer factor and evidence of emphysema on chest computed tomography scans. The lone emphysema data were selected because they might be useful in understanding COPD onset.
References<br>1. Catlett NL, Bargnesi AJ, Ungerer S, Seagaran T, Ladd W, Elliston KO, Pratt D: Reverse causal reasoning: applying qualitative causal knowledge to the interpretation of high-throughput data. BMC bioinformatics 2013, 14:340.<br>2. Chen C, Grennan K, Badner J, Zhang D, Gershon E, Jin L, Liu C: Removing batch effects in analysis of expression microarray data: an evaluation of six batch adjustment methods. PloS one 2011, 6:e17238.

原始网络（Original networks）、NVC网络（NVC networks）及其说明文档。本文件包含已发表的原始网络名称、Bionet网站发布的聚合型NVC网络名称，以及网络说明文档。本次jamboree专题研讨会讨论的15个网络已在"Discussed in Jamboree"列中以“X”标注。
慢性阻塞性肺疾病（Chronic Obstructive Pulmonary Disease，COPD）数据集、对应说明文档，以及第一阶段（Phase 1）构建COPD模型所用的比较分析方案。本研究采用早期COPD患者与健康吸烟者的肺组织、小气道及肺泡巨噬细胞来源的COPD与肺气肿数据集开展反向因果推理分析。构建COPD网络所用的数据集源自基因表达综合数据库（Gene Expression Omnibus，GEO）。
状态变化（state changes，SCs）：通过满足以下标准的差异表达基因定义：错误发现率（False Discovery Rate，FDR）校正后的P值<0.05、倍数变化≥1.3，且Affy平台的最小表达量为100。机制假说（hypotheses，HYPs）：通过SCs与Selventa知识库（Selventa Knowledgebase）[1]预测得到的机制或假说，其截断阈值为：富集度P值<0.1、一致性P值<0.1。
早期COPD定义为慢性阻塞性肺疾病全球倡议（Global Initiative for Chronic Obstructive Lung Disease，GOLD）分期1期与2期。由于每个数据集内早期COPD患者的样本量较小，3个小气道数据集通过ComBat [2]工具进行批次校正合并。
GSE10006数据集中的孤立性肺气肿定义为：肺功能检查正常但转运因子降低，且胸部计算机断层扫描显示存在肺气肿证据的患者。选取孤立性肺气肿数据集是因其有助于阐明COPD的发病机制。
参考文献
1. Catlett NL, Bargnesi AJ, Ungerer S, Seagaran T, Ladd W, Elliston KO, Pratt D: 反向因果推理：将定性因果知识应用于高通量数据解读. BMC Bioinformatics 2013, 14:340.
2. Chen C, Grennan K, Badner J, Zhang D, Gershon E, Jin L, Liu C: 表达微阵列数据分析中的批次效应去除：六种批次校正方法的评估. PLoS ONE 2011, 6:e17238.