DCPR_V1.0

DCPRThis is a repository to deposit the code and data for DCPR model. DCPR is a deep learning framework for circadian phase reconstruction.
Main filesAD Datasets: Three Alzheimer's Disease(AD) datasets from different brain regions: Entorhinal Cortex (EC), Hippocampus (HIP), and Frontal Cortex (FC).
Synthetic datasets: Twelve synthetic datasets differ in varying experimental conditions (e.g., time span, sampling interval, and noise levels).
GEO datasets: Six GEO datasets with diffrent species, tissues and sequencing platforms.
Transcriptomic datasets: Four additional GEO datasets for model test: the transcriptomic profiles of Bmal1 knockout and Cry1/2 knockout mouse model, normal aging, human skeletal muscle.
DCPR_codes: Source code for DCPR model.
Visualization: Scripts for data visualization, including circadian curves (cosine model fitting), sample plots (predicted and real time), model performance (CDF curves, and accuracy plots).
Supplementary File 1: Seed gene sets derived from mouse peritoneal macrophages/hypothalamus and Papio anubis ileum.
Supplementary File 2: Prior knowledge obtained from the Circadian Gene Database (CGDB) and related analysis of core clock genes from mouse liver.
Supplementary File 3: Comparative analysis of differential rhythmic patterns in core clock genes between normal and Alzheimer's disease conditions across three brain region datasets (Entorhinal Cortex (EC), Hippocampus (HIP), and Frontal Cortex (FC)).
ConfigurationIt is recommended to use the conda environment (python 3.8). See environment.yaml for details.
How to userefer to https://github.com/NJAU-CDSIC/DCPR_V1.0

DCPR 本代码库用于存储DCPR模型的配套代码与数据集。DCPR是一款面向昼夜节律相位重构的深度学习框架。
主文件说明：
阿尔茨海默病（Alzheimer's Disease, AD）数据集：包含三个源自不同脑区的阿尔茨海默病数据集，分别为内嗅皮层（Entorhinal Cortex, EC）、海马体（Hippocampus, HIP）与额叶皮层（Frontal Cortex, FC）。
合成数据集：共12个合成数据集，各数据集基于不同实验条件构建，涵盖时间跨度、采样间隔与噪声水平等变量。
GEO数据集：6个GEO数据集，覆盖不同物种、组织类型与测序平台。
转录组数据集：4个额外GEO数据集用于模型验证测试，包括Bmal1基因敲除、Cry1/2基因敲除小鼠模型的转录组谱，正常衰老样本以及人类骨骼肌的转录组特征。
DCPR_codes：DCPR模型的源代码文件。
可视化工具：包含各类数据可视化脚本，可实现昼夜节律曲线（余弦模型拟合）、样本对比绘图（预测时间与真实时间对照）以及模型性能评估（累积分布函数 (Cumulative Distribution Function, CDF) 曲线与准确率绘图）。
补充文件1：源自小鼠腹膜巨噬细胞/下丘脑以及东非狒狒（Papio anubis）回肠的种子基因集。
补充文件2：从节律基因数据库（Circadian Gene Database, CGDB）获取的先验知识，以及小鼠肝脏核心时钟基因的相关分析结果。
补充文件3：针对内嗅皮层（EC）、海马体（HIP）与额叶皮层（FC）三个脑区的数据集，对比分析正常状态与阿尔茨海默病状态下核心时钟基因的差异节律模式。
配置说明：推荐使用conda环境（Python 3.8），详细配置信息请参考environment.yaml文件。
使用方法：请参照链接：https://github.com/NJAU-CDSIC/DCPR_V1.0