The effects of genetic variants primarily occur in differentiated cells meaning we need to access these cell types to measure variant effects for most disease genes. We developed saturation genome editing in stem cells (iPSC-SGE) to enable phenotyping in diverse genetic and cell contexts at scale!
@dougfowler.bsky.social @shawnfayer.bsky.social @dholmes.bsky.social @fritzroth.bsky.social @afrubin.bsky.social @leastarita.bsky.social @wnoble.bsky.social @chanzuckerberg.bsky.social
🤝 Many people to thank, including dougfowler.bsky.social for his mentorship as well as shawnfayer.bsky.social, dholmes.bsky.social, fritzroth.bsky.social, afrubin.bsky.social, leastarita.bsky.social, wnoble.bsky.social and others, and NHGRI and CZI chanzuckerberg.bsky.social��� for funding.
9/9
🔮 Multidimensional variant information may help empower or constrain next generation predictors! Current variant effect predictors perform poorly on molecular and cellular phenotypes, and struggle to parse complex variant-disease relationships.
8/9
UMAP of LMNA variant profiles (left) colored by Louvain-derived clusters. Colors legend at bottom. Graphical representation (right) of lamin A processing, localization, aggregation, and degradation, annotated with their relationship to dimerization, multimerization and variant clusters.
🌎 LMNA variant ➡️ structure ➡️ abundance and localization ➡️ function! VIS-seq maps LMNA variant effects across scales of cellular organization and discovered a new subset of gain-of-function LMNA variants.
7/9
UMAP of iPS cell PTEN variant profiles (left). Triangles indicate association with clinical phenotypes. gnomAD v4.1 (light blue) variants are also plotted. Circles represent other variants, colored green (synonymous) or grey (non-synonymous). Receiver operating characteristic (ROC) curves (right) produced by macro-averaging sensitivity and specificity over classes for models trained on iPS and neuron VIS-seq landmark features (dotted lines) as well as scores from (C) classifying gnomAD controls from PHTS-associated variants from ASD/DD-associated variants. AUC is shown on the right for each model.
📊 PTEN variants ≠ one axis of “function”. VIS-seq’s multidimensional representations discriminate between PTEN autism- and tumor syndrome-associated variants.
6/9
Images of libraries of mEGFP-tagged lamin A in U2OS cells (left), mEGFP-tagged PTEN in human WTC11 PTEN-KO inducible-NGN2 iPS cells or mEGFP-tagged H1.4 and RPS19 in human WTC11 iPS cells (right), or mEGFP-tagged PTEN library in derived neurons, seven days after induction of NGN2 (far right). Some cells express variants with localization defects (lamin A=nuclear aggregates, PTEN=nucleocytoplasmic, HIST1H1E=chromatin binding, RPS19=nucleolar-cytoplasmic). Scale bar indicates 20 μm.
🌐 Generalizability and scale: ~3000 LMNA or PTEN variants • 11.4 million cells • 1000+ image-derived features per cell • Fluorescent proteins, antibodies, and RNA FISH readouts • Cancer cell lines, iPS and derived cells.
5/9
VIS-seq uses fluorescent in situ sequencing of abundant circular RNA barcodes to genotype cells expressing protein variants. (1) A variant library in the VIS-seq expression cassette is integrated into cells via piggyBac-ase. (2) Cells are fixed; barcodes are reverse transcribed, captured with a padlock probe and amplified; (3) cells are stained and imaged; (4) barcode is sequenced in situ; (5) single cell phenotype-genotype pairs are determined using STARCall; and (6) features for each cell are extracted using CellProfiler.
🎯 Barcoded circular RNAs are co-expressed in each cell along with a tagged variant ➡️ pooled imaging ➡️ in situ sequencing decodes the barcode ➡️ CellProfiler extracts thousands of molecular & cellular features per cell!
4/9
🤿 Dive-in yourself at visseq.gs.washington.edu! A website showing profiles, features, and cell images of LMNA and PTEN variants built by lab member and recent UW Computer Science graduate Nick Bradley.
3/9
⚡ I developed VIS-seq with the help of dougfowler.bsky.social and others at UW Genome Sciences. Check out my preprint:
www.biorxiv.org/content/10.1...
2/9
🚨 Most variant screens measure growth or abundance. What do they miss? That variants impact a spectrum of protein and cellular phenotypes. Variant in situ sequencing (VIS-seq) finds what’s missing: image cells 🔬 first, decode later, revealing multi-scale phenotypes for thousands of variants.👇
1/9