Supplemental Tables S3.3, S3.4, S3.5, and S3.6

TABLE S3.3. Lists of all DEGs, Enriched GO terms, and KEGG
Pathways Identified as Differentially Expressed in at Least One Comparison. Analysis
was completed with Blast2GO.

TABLE S3.4. A List of the Top 250 Genes with Greatest Relative
Importance in Predicting an Environmental Variable with a Specified Artificial
Neural Network Model. If a gene is included in a variable’s list of top 250, it
is marked with an “x”. The R values were calculated with Pearson’s correlation
using the expected and actual values following ANN training. The SSE values
represent the model’s sum of squares error. The overall top 250 genes across
all samples were selected for further ANN analysis.

TABLE S3.5. The Relative Importance of 250 O. Lurida
genes in Predicting an Environmental Variable with a Specified Artificial
Neural Network Model. The R values were calculated with Pearson’s correlation
using the expected and actual values following ANN training. The SSE values
represent the model’s sum of squares error.

TABLE S3.6. The Relative Importance of Each
Environmental Variable in Predicting the Gene Expression of 1920 O. Lurida
Differentially Expressed Genes Through a Specified Artificial Neural Network Model.
The R values were calculated with Pearson’s correlation using the expected and
actual values following ANN training. The SSE values represent the model’s sum
of squares error.

表S3.3：至少在一组比较中被鉴定为差异表达基因（DEGs, Differentially Expressed Genes）、富集基因本体（GO, Gene Ontology）术语及京都基因与基因组百科全书（KEGG, Kyoto Encyclopedia of Genes and Genomes）通路的完整列表。本分析通过Blast2GO完成。

表S3.4：利用指定人工神经网络（ANN, Artificial Neural Network）模型预测环境变量时，相对重要性排名前250的基因列表。若某基因被纳入某变量的前250基因列表，则以"x"标注。R值通过皮尔逊相关系数计算，所用数据为人工神经网络训练后的预测值与实际观测值。SSE值代表模型的残差平方和。针对所有样本中综合排名前250的基因，将其用于后续人工神经网络分析。

表S3.5：利用指定人工神经网络（ANN）模型预测环境变量时，250个O. Lurida基因的相对重要性。R值通过皮尔逊相关系数计算，所用数据为人工神经网络训练后的预测值与实际观测值。SSE值代表模型的残差平方和。

表S3.6：利用指定人工神经网络（ANN）模型预测1920个O. Lurida差异表达基因的基因表达量时，各环境变量的相对重要性。R值通过皮尔逊相关系数计算，所用数据为人工神经网络训练后的预测值与实际观测值。SSE值代表模型的残差平方和。