CASowary is a command line machine learning tool written in Python 3.6.8 for predicting the efficacy of sgRNAs for CRISPR Cas13 and other transcript targeting systems.

Software and library requirements

Python libraries:

argparse (version 1.1)
numpy (version 1.16.2)
os (3.6)
random (3.6)
sklearn (version 0.20.3)

Other software:

TopHat aligner (version 2.1.1)
bedtools (version 2.26.0)

Repository:

https://github.com/Janga-Lab/CASowary.git

Files included: Gene information (Gene_Information.txt), Reference transcriptome file (hg38_transcriptome.fasta), Cell line protein occupancy file (Hek293_Occupancy_Profile.xls, Hela_Occupancy_Profile.xls), Gene input file (Gene.csv), List of k-mers (Gini_K-mers.txt), and training data for the model (Training_Data_Gini.txt)

Note: Due to the size of the Gene_Information.txt and hg38_transcriptome.fasta they are no longer available on GitHub. You can download them at the following link: https://indiana-my.sharepoint.com/:f:/g/personal/amkrug_iu_edu/Et2HtXXEQLBDpFzbM_cbDc4B2yiQwJEK9YiJeEOOJYOPIA?e=PbUGCt

How it works:

CASowary has two different scripts that must be run in a particular order:

a. CASowary_Preprocess - which takes a list of gene names as input to create a fasta file of all possible guides that tile all transcripts associated with that gene.

b. CASowary_Predict - which takes the aforementioned fasta file, the TopHat alignment file, and a protein occupancy peak file to create guide predictions.

CASowary_Preprocess can be run using the following command.

python CASowary_Preprocess.py Gene.csv -o Input_Guides.fasta

Intermediate steps:

The resulting Input_Guides.fasta from CASowary_Preprocess script, needs to be run through the TopHat aligner with the following parameters:

read-mismatches 3, read-edit-dist 3 (see example below).

tophat --read-mismatches 3 --read-edit-dist 3 -o Gene_Guides ./BT_hg38_transcriptome_0221/BT_hg38_transcriptome_0221 Input_Guides.fasta

Note: The reference transcriptiome file should be stored at the same working directory or link with the Gene_Information.txt file.

The results of the TopHat alignment (.bam) files, can be converted into bed file using bedtools 'bamtobed' command (see example below).

bedtools bamtobed -i accepted_hits.bam > accepted_hits.bed

The resulting bed file should also be sorted using the bedtools 'sort' command (see example below).

bedtools sort -i Target_Hits.bed > Target_Hits.sorted.bed

Finally, CASowary_Predict script can be run using the following command.

python CASowary_Predict.py Input_Guides.fasta -t Target_Hits.sorted.bed -p Hek293_Occupancy_Profile.xls -o Predictions.txt

File description

Gene.csv is a csv file containing the input gene names of interest for CASowary_Preprocess

Input_Guides.fasta is the fasta file generated by CASowary_Preprocess.

Target_Hits.sorted.bed is the sorted bed file from the tophat alignement.

Hek293_Occupancy_Profile.xls is the peak file for protein occupied regions in Hek293 cell line.

Predictions.txt is a tab separated output file containing the guide name, transcript name, guide sequence, model predction, and the model's confidence in that prediction.