Back

KaryoTap Enables Aneuploidy Detection in Thousands of Single Human Cells

Mays, J. C.; Mei, S.; Bosco, N.; Zhao, X.; Bianchi, J. J.; Kidiyoor, G. R.; Holt, L. J.; Davoli, T.

2023-09-10 genomics
10.1101/2023.09.08.555746 bioRxiv
Show abstract

Aneuploidy--the presence of chromosome gains and losses--is highly prevalent in human tumors, yet the contribution of chromosome missegregation rates and selection to shape these patterns remains poorly understood. To address this challenge at scale, we developed KaryoTap, a cost-effective single-cell DNA sequencing method combining custom targeted panels for the Tapestri platform with a Bayesian gaussian mixture model to detect chromosome- and chromosome arm-scale aneuploidy as well as integrated gRNAs and barcodes. KaryoTap achieves an average accuracy of >85% for arm events and >90% for chromosome events at <$1 per cell, allowing scalable analysis of tens of thousands of cells per experiment. Through KaryoTap-based analysis of 11,555 cells, we performed in vitro evolution screens on immortalized human cells from mammary gland, pancreas and melanocytes. By comparing aneuploidy frequencies immediately after reversine-induced missegregation versus after extended proliferation, we quantified positive and negative selection for specific aneuploidies. Most aneuploidies--both gains and losses--are under negative selection, yet some chromosomal gains are under positive selection (such as 8q and 7q). We found that proliferative selective pressures can explain tissue-specific patterns of chromosomal gains observed in human cancers. Critically, when positively selected events were excluded from analysis, correlations between in vitro and cancer gain frequencies strongly decreased or disappeared, whereas excluding negatively selected events largely preserved these correlations. These findings demonstrate that proliferative selection shapes the landscape of chromosomal gains in cancer, with a more prominent role for positive selection.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.