Back

An unsupervised deep learning framework encodes super-resolved image features to decode bacterial cell cycle

Griffie, J.; Zhang, C.; Denereaz, J.; Pham, T.-A. M.; Weissbart, G.; Sieben, C.; Lambert, A.; Veening, J.-W.; Manley, S.

2024-03-14 bioinformatics
10.1101/2024.03.13.584804 bioRxiv
Show abstract

Super-resolution microscopy can resolve cellular features at the nanoscale. However, increased spatial resolution comes with increased phototoxicity, and reduced temporal resolution. As a result, studies that require the highest spatial resolutions often rely on static or fixed images, lacking dynamic information. This is particularly true of bacteria, whose lateral dimensions approach the scale of the diffraction limit. In this work, we present Enso, a method based on unsupervised machine learning to recover bacterial cell cycle and cell type information from static single molecule localization microscopy (SMLM) images, whilst retaining nanoscale spatial resolution. Enso uses single-cell images as input, and orders cells according to their spatial pattern progression, ultimately linked to the cell cycle. Our method requires no a priori knowledge or categories, and is validated on both simulated and user-annotated experimental data.

Matching journals

The top 2 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.