Clone the repo!
git clone https://github.com/pvtodorov/rotation_tune.git
Make the batch script executable.
cd rotation_tune
chmod 700 Scripts/run_pipeline.sh
Run the scripts
./Scripts/run_pipeline.sh
Scripts/get_sample_ids_from_meta.R ingests a YAML file.
Here studies can be specified as follows:
-
Sample_series_id: GSE98519
-
Sample_series_id: GSE100599
Sample_source_name_ch1: hypothalamus
-
Sample_series_id: GSE98356
Sample_source_name_ch1: ['Hypothalamus_ad libitum (AL)',
'Hypothalamus_ad libitum after diet restricted',
'Hypothalamus_diet restricted']
These filter the metadata according to the specified key and values.
| Script | What it does |
|---|---|
| run_pipeline.sh | Creates Outputs/ and Data/ directories if they do not exist. If the Data/ directory and thus the ARCHS4 mouse_matrix.h5 is not found, it will be downloaded using wget. |
| get_sample_ids_from_meta.R | Given a path to the YAML configuration it loads the ARCHS4 h5 file and selects the studies and samples from it. The script then dumps a YAML file keyed to GSE with a list of indicies for the samples within the matrix. |
| get_gene_idxs.R | Given a path to a YAML containing sample indicies and a per study and all study threshold it writes to a YAML file keyed to GSE with genes that are non-zero for n samples specified per study and all study threshold. |
| WGCNA_outlier_removal.R | Takes in the the sample and gene indicies and applies per-study WGCNA outlier detection as performed by Gandall on arrays here. Dumps a YAML keyed to GSEs with sample indicies as values but with the outliers excluded. |
| batch_correction.R | load all samples for all studies, performs batch correction using GSEs as batches, saves a CSV file with samples as columns and genes as rows. |