Model Overview

Section 1

Model Overview

CELL-E 2 is the second iteration of the original CELL-E model which utilizes an amino acid sequence and nucleus image to make predictions of subcellular protein localization with respect to the nucleus.

We use a novel bidirectional transformer that can generate images depicting protein subcellular localization from the amino acid sequences (and vice versa).

CELL-E 2 can not only capture the spatial complexity of protein localization and produce probability estimates of localization atop a nucleus image, but it is also able to generate sequences from images, enabling de novo protein design.

We trained on the Human Protein Atlas (HPA) and the OpenCell datasets. CELL-E 2 utilizes pretrained amino acid embeddings from ESM-2. Localization is predicted as a binary image atop the provided nucleus. The logit values are weighted against these binary images to produce a heatmap of expected localization.

Section 2

Localization Prediction

Text-to-Image

CELL-E 2 can generate localization images by masking the image input section.

example images: nucleus, protein

Section 3

Sequence Prediction

Image-to-Text

Similarly, amino acids positions can be masked (replaced or inserted) to make predictions based on the localization pattern.

example images: nucleus, protein

Section 4

De novo Protein Design

We created an entirely new approach to protein design which leverages spatial information from images.

Using CELL-E 2, we predicted 255 likely novel nuclear localizing signals with distinct sequence homology from documented sequences.

Section 5

Comparison

In comparison to CELL-E, CELL-E 2 makes image predictions 65x faster with higher accuracy.