CELL-E 2: Translating Proteins to Pictures and Back with a Bidirectional Text-to-Image Transformer

Overview

Journal Adv Neural Inf Process Syst

Date 2024 Jul 18

PMID 39021511

Authors

Emaad Khwaja

Yun S Song

Aaron Agarunov

Bo Huang

Affiliations

Soon will be listed here.

Abstract

We present CELL-E 2, a novel bidirectional transformer that can generate images depicting protein subcellular localization from the amino acid sequences (and ). Protein localization is a challenging problem that requires integrating sequence and image information, which most existing methods ignore. CELL-E 2 extends the work of CELL-E, not only capturing the spatial complexity of protein localization and produce probability estimates of localization atop a nucleus image, but also being able to generate sequences from images, enabling protein design. We train and finetune CELL-E 2 on two large-scale datasets of human proteins. We also demonstrate how to use CELL-E 2 to create hundreds of novel nuclear localization signals (NLS). Results and interactive demos are featured at https://bohuanglab.github.io/CELL-E_2/.

References

. IUPAC-IUB Commission on Biochemical Nomenclature. A one-letter notation for amino acid sequences. Tentative rules. J Biol Chem. 1968; 243(13):3557-9. View

Lange A, Mills R, Lange C, Stewart M, Devine S, Corbett A . Classical nuclear localization signals: definition, function, and interaction with importin alpha. J Biol Chem. 2006; 282(8):5101-5. PMC: 4502416. DOI: 10.1074/jbc.R600026200. View

Bradley K, Bowl M, Williams S, Ahmad B, Partridge C, Patmanidi A . Parafibromin is a nuclear protein with a functional monopartite nuclear localization signal. Oncogene. 2006; 26(8):1213-21. DOI: 10.1038/sj.onc.1209893. View

Thul P, Lindskog C . The human protein atlas: A spatial map of the human proteome. Protein Sci. 2017; 27(1):233-244. PMC: 5734309. DOI: 10.1002/pro.3307. View

Schnell U, Dijk F, Sjollema K, Giepmans B . Immunolabeling artifacts and the need for live-cell imaging. Nat Methods. 2012; 9(2):152-8. DOI: 10.1038/nmeth.1855. View

Bhardwaj G, OConnor J, Rettie S, Huang Y, Ramelot T, Mulligan V . Accurate de novo design of membrane-traversing macrocycles. Cell. 2022; 185(19):3520-3532.e26. PMC: 9490236. DOI: 10.1016/j.cell.2022.07.019. View

Nguyen Ba A, Pogoutse A, Provart N, Moses A . NLStradamus: a simple Hidden Markov Model for nuclear localization signal prediction. BMC Bioinformatics. 2009; 10:202. PMC: 2711084. DOI: 10.1186/1471-2105-10-202. View

Lu J, Wu T, Zhang B, Liu S, Song W, Qiao J . Types of nuclear localization signals and mechanisms of protein import into the nucleus. Cell Commun Signal. 2021; 19(1):60. PMC: 8140498. DOI: 10.1186/s12964-021-00741-y. View

Thumuluri V, Almagro Armenteros J, Johansen A, Nielsen H, Winther O . DeepLoc 2.0: multi-label subcellular localization prediction using protein language models. Nucleic Acids Res. 2022; 50(W1):W228-W234. PMC: 9252801. DOI: 10.1093/nar/gkac278. View

10.

Guo Y, Yang Y, Huang Y, Shen H . Discovering nuclear targeting signal sequence through protein language learning and multivariate analysis. Anal Biochem. 2019; 591:113565. DOI: 10.1016/j.ab.2019.113565. View

11.

Schnoes A, Ream D, Thorman A, Babbitt P, Friedberg I . Biases in the experimental annotations of protein function and their effect on our understanding of protein function space. PLoS Comput Biol. 2013; 9(5):e1003063. PMC: 3667760. DOI: 10.1371/journal.pcbi.1003063. View

12.

Seibel N, Eljouni J, Nalaskowski M, Hampe W . Nuclear localization of enhanced green fluorescent protein homomultimers. Anal Biochem. 2007; 368(1):95-9. DOI: 10.1016/j.ab.2007.05.025. View

13.

von Mering C, Krause R, Snel B, Cornell M, Oliver S, Fields S . Comparative assessment of large-scale data sets of protein-protein interactions. Nature. 2002; 417(6887):399-403. DOI: 10.1038/nature750. View

14.

Lin Z, Akin H, Rao R, Hie B, Zhu Z, Lu W . Evolutionary-scale prediction of atomic-level protein structure with a language model. Science. 2023; 379(6637):1123-1130. DOI: 10.1126/science.ade2574. View

15.

Jiang Y, Wang D, Yao Y, Eubel H, Kunzler P, Moller I . MULocDeep: A deep-learning framework for protein subcellular and suborganellar localization prediction with residue-level interpretation. Comput Struct Biotechnol J. 2021; 19:4825-4839. PMC: 8426535. DOI: 10.1016/j.csbj.2021.08.027. View

16.

Nielsen H, Tsirigos K, Brunak S, von Heijne G . A Brief History of Protein Sorting Prediction. Protein J. 2019; 38(3):200-216. PMC: 6589146. DOI: 10.1007/s10930-019-09838-3. View

17.

Parnamaa T, Parts L . Accurate Classification of Protein Subcellular Localization from High-Throughput Microscopy Images Using Deep Learning. G3 (Bethesda). 2017; 7(5):1385-1392. PMC: 5427497. DOI: 10.1534/g3.116.033654. View

18.

Rao R, Bhattacharya N, Thomas N, Duan Y, Chen X, Canny J . Evaluating Protein Transfer Learning with TAPE. Adv Neural Inf Process Syst. 2021; 32:9689-9701. PMC: 7774645. View

19.

Kitamura A, Nakayama Y, Kinjo M . Efficient and dynamic nuclear localization of green fluorescent protein via RNA binding. Biochem Biophys Res Commun. 2015; 463(3):401-6. DOI: 10.1016/j.bbrc.2015.05.084. View

20.

Madani A, Krause B, Greene E, Subramanian S, Mohr B, Holton J . Large language models generate functional protein sequences across diverse families. Nat Biotechnol. 2023; 41(8):1099-1106. PMC: 10400306. DOI: 10.1038/s41587-022-01618-2. View