Orobas

Orobas is an R package (with Python modules) for scoring chemical-genetic CRISPR screening data.

Overview

Orobas is a computational approach for transforming raw data from CRISPR-Cas9 chemical-genetic screens into quantitative interaction scores. Orobas computes differential interaction scores with accompanying statistical tests that account for multiple CRISPR guides per gene. The method includes approaches for post-processing differential log2 fold-change scores across multiple screens, incorporating normalization to reduce technical artifacts and correct batch effects.

Installation

1. Install a Virtual Environment Manager

We recommend installing Anaconda or any other virtual environment manager of your choice.

---

2. Clone the Repository

Download the Orobas source code:

git clone https://github.com/csbio/orobas.git
cd orobas

---

3. Create the Environment

Create a virtual environment using the provided YAML configuration:

conda env create -f orobas_environment.yml

By default, this will create an environment named:

orobas_env

You can change the environment name by editing the name: field in orobas_environment.yml.

---

4. Activate the Environment

Activate the environment before running Orobas modules:

conda activate orobas_env

---

Note:

• Recommended versions:
  • Python >= 3.9
  • R >= 3.6
• The code in this protocol was executed using:
  • Python 3.9
  • R 4.4.3

---

How to run

After activating the environment, you can run Orobas as described in the example code and the protocol.

---

Expected Outputs

Output directory and files from single-screen scoring:

.
├── <output>
│   ├── <screen-batch-1>    # (Directory)                  
│   │   ├── ... 
│   ├── <screen-batch-2>    # (Directory)
│   │   ├── qc    # (Directory) screen replicate LFC scatter plots and other quality control files                                       
│   │   │   ├── essential_PR_QC.tsv    # precision-recall AUC of essential-targeting guides from screen replicates
│   │   │   ├── lfc_heatmap.png    # heatmap of Pearson Correlation among screen replicate LFCs
│   │   │   ├── replicate_cor.tsv    # Pearson Correlation among screen replicate LFCs
│   │   │   ├── <screen-batch-2>_<condition-screen-1-replicate-A>_vs_<screen-batch-2>_<condition-screen-1-replicate-B>_replicate_comparison.png
│   │   │   ├── ...
│   │   │   ├── <screen-batch-2>_<control-screen-1-replicate-A>_vs_<screen-batch-2>_<control-screen-1-replicate-B>_replicate_comparison.png
│   │   │   ├── ...
│   │   │   ├── reads    # (Directory) raw read count histograms of screen replicates and other quality control files
│   │   │   │   ├── total_reads.png    # bar-plot of raw read counts from all screen replicates   
│   │   │   │   ├── reads_heatmap.png    # heatmap of Pearson Correlation among screen replicate raw read counts
│   │   │   │   ├── <screen-batch-2_T0>_raw_reads_histogram.png
│   │   │   │   ├── ...
│   │   │   │   ├── <screen-batch-2>_<control-screen-1-replicate-A>_raw_reads_histogram.png
│   │   │   │   ├── ...
│   │   │   │   ├── <screen-batch-2>_<condition-screen-1-replicate-A>_raw_reads_histogram.png
│   │   │   │   ├── ...
│   │   ├── guide_dlfc    # (Directory) guide-level replicate-level dLFC score file
│   │   │   ├── <screen-batch-2>_<condition-screen-1>_vs_<screen-batch-2>_<control-screen-1>_guide_dlfc_pre_jk.tsv
│   │   │   ├── ...
│   │   ├── plots    # (Directory) scatter plots of gene-level condition LFCs vs control LFCs with negative and positive interactions
│   │   │   ├── <screen-batch-2>_<condition-screen-1>_vs_<screen-batch-2>_<control-screen-1>_scatter.png
│   │   │   ├── ...
│   │   ├── condition_gene_calls.tsv    # ***score file containing gene-level screen-level LFC, dLFC, FDR, significant hits and other values
│   │   ├── t0_normalized_screens_guide_level.tsv    # guide-level replicate-level LFC score file 
│   ├──  <screen-batch-3>    # (Directory)
│   │   ├── ... 
│   ├── ... 
│   ├── differential_LFC_scores.tsv    # gene-level dLFC scores from all screens from all screen-batches 
│   ├── fdr_scores.tsv    # gene-level FDR scores from all screens from all screen-batches

Output directory organization and files from global-normalization:

.
├── global_normalization    # (Directory) output files generated after running global normalization
│   ├── global_normalized_dLFC_scores.tsv    # ***file with normalized dLFC scores from all selected condition screens                      
│   ├── fdr_scores_all.tsv    # file with FDR scores from all selected condition screens
│   ├── scores_all.csv    # file with LFC, normalized dLFC, FDR scores, and updated significant hits from all selected condition screens
│   ├── wbc_scores.csv    # file with within-between correlation scores after each normalization step
│   ├── sd_scale_table.tsv    # standard deviation of dLFC scores before and after scaling step
│   ├── control    # (Directory) control screen files
│   │   ├── control    # (Directory) control dLFC score file
│   │   │   ├── control_effect_scores.tsv    # file with dLFC scores from control screens
│   │   ├── control_control_map_table.tsv    # control_control_map table used in generating control dLFC scores
│   │   ├── control_replicates_map_table.tsv    # control_replicates_map table used in generating control dLFC scores
│   │   ├── replicate_cor.tsv    # Pearson correlation among control screen replicates
│   ├── LDA_evaluation_plots    # (Directory) global ROCAUC and per-screen ROCAUC histograms at each LDA component removal step
│   │   ├── bc_lda_<component_number>_histogram.png
│   │   ├── ...
│   │   ├── bc_lda_<component_number>_roc.png
│   │   ├── ...
│   ├── plots    # (Directory) scatter plots of gene-level condition LFCs vs control LFCs with negative and positive interactions
│   │   ├── <screen-batch-1>_<condition-screen-1>_vs_<screen-batch-1>_<control-screen-1>_scatter.png
│   │   ├── <screen-batch-1>_<condition-screen-2>_vs_<screen-batch-1>_<control-screen-2>_scatter.png
│   │   ├── <screen-batch-2>_<condition-screen-1>_vs_<screen-batch-2>_<control-screen-1>_scatter.png
│   │   ├── ...

Citation

If you use Orobas in your work, please cite:

[Publication placeholder — coming soon]

---

License

This project is distributed under the MIT License.