This repository contains the code used in the paper:
**A CRISPRi screen in E. coli reveals an unexpected sequence-specific toxicity of dCas9**
Lun Cui, Antoine Vigouroux, Francois Rousset, Hugo Varet, Varun Khanna & David Bikard
In this study We designed a library of 84215 unique guide RNAs targeting random positions
along the genome of E. coli MG1655, with the simple requirement of a “NGG” PAM. The library
contains an average of 19 targets per gene. A pool of guide RNAs obtained through on-chip
oligo synthesis was cloned under the control of a constitutive promoter on plasmid psgRNA
and electroporated in strain LC-E18 carrying the dCas9 gene under the control of a Ptet
promoter in the chromosome. The pooled library of cells was then grown in rich medium
over 17 generations with anhydrotetracycline (aTc). The effect of each guide on the cell fitness can be measured as
the fold change in abundance (log2FC) of the guide RNA in the library
during the course of the experiment, as measured through deep sequencing of the library.
The log2FC data for all guides in the screen can be found in the file Supplementary_table_6-screen_data.csv
The analysis is divded in three jupyter notebooks:
* Effect of dCas9 binding position and orientation.ipynb
> In this notebook we look at the effect of dCas9 binding position and orientation. In particular we perform an analysis of polar and reverse-polar effects.
* off-target analysis.ipynb
> In this notebook we investigate whether the unexpected fitness defect produced by some guides in the screen can be explained by off-target positions blocking the expression of essential or fitness genes.
* bad-seed analysis.ipynb
> In this notebook we use a machine learning approach to reveal how dCas9 can kill _E. coli_ when guided by some specific 5nt PAM-proximal sequences
