Childs & Morabito et al. Cell Reports 2024
This Repository contains the code used for data processing and analysis in our manuscript titled "Relapse to cocaine-seeking is regulated by medial habenula Nr4a2 in mice". In this study we used single-nucleus RNA-seq (snRNA-seq) to profile the habenula in four different groups of mice to study the molecular changes following a cocaine reinstatement behavioral experiment and manipulation of the transcription factor Nr4a2. Here we list the major sections of the paper and provide links to the data analysis steps in each part.
The raw and processed snRNA-seq data (Seurat format) generated in this study has been deposited on the NCBI Gene Expression Omnibus (GEO) at accession number GSE208081.
We supply the script that was used during the cocaine reinstatement behavioral experiment.
snRNA-seq was performed using the 10X Genomics kit, and we used CellRanger to quantify gene expression from the raw sequencing reads. After running CellRanger, ambient RNA was removed using cellbender.
In our snRNA-seq dataset we had four different groups of mice: NURR2C and GFP mice that were behaviorally experienced and behaviorally naive. The behaviorally naive and behaviorally experienced groups were analyzed separately before performing an integrated analysis.
- Behaviorally naive clustering analysis (Jupyter Notebook)
- Behaviorally experienced clustering analysis (Jupyter Notebook)
- Integrated clustering analysis (Jupyter Notebook)
- Converting anndata to Seurat object (R markdown)
- Plotting clustering results and QC metrics (R markdown)
- Cluster marker gene script (bash script)
- Cell type marker gene script (bash script)
- Misc. plotting utility script (R script)
To compare our snRNA-seq with a previously published dataset of the mouse habenula, we performed an additional integrated analysis.
Due to our experimental manipulation of Nr4a2 in the habenula, we were interested in investigating snRNA-seq read pileup at the Nr4a2 locus. For this analysis, we processed the sequencing data so that it could be viewed on the genome browser.
- Script to make genome browser trackhubs (R markdown)
- Alternative splicing analysis with Swan (Jupyter Notebook)
- Additional helper scripts are in this directory
As part of this study, we developed a custom strategy for TF network analysis. These functions have been formally added to the hdWGCNA R package, but we provide the code as is here from before these functions were incuded in hdWGCNA.
- TF network analysis (R markdown)
- Individual functions to construct the TF nets (R script)
- Script to run TF net construction (R script)
- Script to launch jobs to build separate TF nets on the HPC (bash script)
We performed differential expression analysis in each cell type and cell cluster to compare gene expression signatures between behaviorally naive (Figure 4) and behaviorally experienced (Figure 5) NURR2C and GFP mice. We also compared these DEGs to each other (Figure 5). Note that the results plotting R markdown file contains plotting code for Figures 4 and 5, which summarize the DEG results for the naive and experienced mice.
Behaviorally naive (Figure 4)
Behaviorally experienced (Figure 5)
We also performed relative likelihood analysis with MELD in the behaviorally experienced mice to quantify the transcriptome-wide perturbation effects of Nr4a2 manipulation.
We used hdWGCNA to perform co-expression network analysis to specifically study systems-level transcriptome changes in the medial habenula neurons, and to link TF regulatory networks to co-expression modules.