The data presented in this web portal are from pooled, image-based CRISPR knockout screens in HeLa cells. First, a fixed-cell primary screen targeted 5072 fitness-conferring genes, with phenotypes measured using widefield fluorescence microscopy of stains for DNA (DAPI), tubulin (antibody), gamma-H2AX (antibody), and actin (phalloidin). A secondary screen collected live-cell image phenotypes of cells expressing mCherry-tagged Histone H2B for 239 of these gene targets. The HeLa cells used in these screens have doxycycline-inducible Cas9 expression for precise control of the timing between knockout and phenotype measurement for these fitness-conferring genes. For the primary screen, cells were fixed 78 hours after Cas9 expression induction. In the live-cell secondary screen, we tracked cells for 24 hours (10 minute time resolution) for two separate time courses: 48 hours -> 71 hours post-Cas9 induction (referred to as “Day 3”) and 72 hours-> 96 hours post-Cas9 induction (“Day 4”). The pooled knockouts were demultiplexed in situ and matched to the individual cell phenotype images using the pooled optical pooled screening approach (see Feldman, et al. ).
This project is a collaboration between the Cheeseman Lab at the Whitehead Institute and the Blainey Lab at the Broad Institute, with special thanks to Andy Nutter-Upham and Scott McCallum at the Whitehead Institute for their generous work in developing this web portal.
Cell image montages
The cells displayed in each montage were randomly selected from all
acquired images for each sgRNA. For the primary screen, up to 100
cells are displayed for each sgRNA (median of 1,454 acquired in total
for each sgRNA). Separate secondary screen montages are included for
the day 3 and day 4 time courses, with up to 150 mitotic events
included for each sgRNA (median of 290 acquired in total across
sgRNAs and timepoints). These timelapse montages were computationally
aligned such that all cells enter mitosis at approximately the same
frame, and are in sorted order by mitotic event duration. Timestamps
denote time relative to approximate mitotic entry, and the 10 frames
before mitotic entry and after mitotic exit are included in the
montage if available (i.e., are within the acquired time course).
Fixed-cell primary screen features
The fixed-cell measurements included in this data portal are a small
subset of the 1084 total phenotype features extracted from each cell
image, chosen based on their combined ability to discriminate between
functional phenotypes and for their relatively simple interpretation.
The values reported here are gene-level summary scores for each
feature. These scores are computed by taking the median of values
across individual cells expressing the same sgRNA, and then taking
the median of these values among sgRNAs targeting the same gene.
Feature values in this portal are also normalized, meaning the values
are standardized by the median and median absolute deviation of cells
expressing non-targeting sgRNAs within the same well, prior to
summary scoring. Reported FDR q-values were computed by comparing
summary scores to null distributions of corresponding bootstrapped
summary scores from cells expressing non-targeting guides to generate
raw p-values (bootstrap sample size matching the cell and sgRNA
sample size of each gene target, 100,000 bootstrap repetitions each
for cell and sgRNA levels), and then applying the Benjamini-Hochberg
procedure to obtain FDR bounds.
The features are defined in hierarchical form: