Self-Supervised Contrastive Learning of Protein Representations By Mutual Information Maximization
Open accessThe DOI matches its Crossref record; 25 of the 25 references it cites were found in a registry or an index.
Abstract
Pretrained embedding representations of biological sequences which capture meaningful properties can alleviate many problems associated with supervised learning in biology. We apply the principle of mutual information maximization between local and global information as a self-supervised pretraining signal for protein embeddings. To do so, we divide protein sequences into fixed size fragments, and train an autoregressive model to distinguish between subsequent fragments from the same protein and fragments from random proteins. Our model, CPCProt, achieves comparable performance to state-of-the-art self-supervised models for protein sequence embeddings on various downstream tasks, but reduces the number of parameters down to 2% to 10% of benchmarked models. Further, we explore how downstream assessment protocols affect embedding evaluation, and the effect of contrastive learning hyperparameters on empirical performance. We hope that these results will inform the development of contrastive learning methods in protein biology and other modalities.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
UniProt: a worldwide hub of protein knowledge
The UniProt Consortium (2019) · Nucleic Acids Research · doi:10.1093/nar/gky1049 · Cited by 7,698 · cited 1×
“The cost and time associated with obtaining labels for supervised problems on proteins is a challenge, rendering self-supervised methods for obtaining labels an appealing solu-tion.”
BackgroundRepresentation Learning with Contrastive Predictive Coding
Oord, A. V. D., Li, Y., & Vinyals, O. (2018) · arXiv · doi:10.48550/arxiv.1807.03748 · Cited by 6,242
Context Encoders: Feature Learning by Inpainting
Pathak, D., Krahenbuhl, P., Donahue, J., Darrell, T., & Efros, A. A. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.278 · Cited by 5,691
The Pfam protein families database in 2019
El-Gebali, S., Mistry, J., Bateman, A., Eddy, S. R., Luciani, A., Potter, S. C., Qureshi, M., Richardson, L. J., Salazar, G. A., Smart, A., et al. (2019) · Nucleic Acids Research · doi:10.1093/nar/gky995 · Cited by 4,602 · cited 1×
“Additional Pretraining Details Models are pretrained on protein domain sequences from the Pfam database [10] us-ing the same data splits as used for downstream benchmarks [29].”
MethodUnsupervised Feature Learning via Non-parametric Instance Discrimination
Wu, Z., Xiong, Y., Yu, S. X., & Lin, D. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00393 · Cited by 3,664
Rives, A., Meier, J., Sercu, T., Goyal, S., Lin, Z., Liu, J., Guo, D., Ott, M., Zitnick, C. L., Ma, J., et al. (2021) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2016239118 · Cited by 3,286
Rives, A., Meier, J., Sercu, T., Goyal, S., Lin, Z., Liu, J., Guo, D., Ott, M., Zitnick, C. L., Ma, J., et al. (2019) · doi:10.1101/622803 · Cited by 3,280 · cited 1×
“…many of these works directly take embedding techniques from natural language processing (NLP) tasks (Yang et al., 2018a; Riesselman et al., 2019; Rives et al., 2019; Alley et al., 2019; Heinzinger et al., 2019; Elnaggar et al., 2019; Armenteros et al., 2019; Madani et al., 2020; Elnag-gar et…”
BackgroundUnified rational protein engineering with sequence-based deep representation learning
Alley, E. C., Khimulya, G., Biswas, S., AlQuraishi, M., & Church, G. M. (2019) · Nature Methods · doi:10.1038/s41592-019-0598-1 · Cited by 1,131
Data-Efficient Image Recognition with Contrastive Predictive Coding
Hénaff, O. J., Srinivas, A., De Fauw, J., Razavi, A., Doersch, C., Eslami, S. M. A., & Oord, A. V. D. (2019) · arXiv · doi:10.48550/arxiv.1905.09272 · Cited by 1,014
Wang, T., & Isola, P. (2020) · arXiv · doi:10.48550/arxiv.2005.10242 · Cited by 921
Do ImageNet Classifiers Generalize to ImageNet?
Recht, B., Roelofs, R., Schmidt, L., & Shankar, V. (2019) · arXiv · doi:10.48550/arxiv.1902.10811 · Cited by 790
What makes for good views for contrastive learning
Tian, Y., Sun, C., Poole, B., Krishnan, D., Schmid, C., & Isola, P. (2020) · arXiv · doi:10.48550/arxiv.2005.10243 · Cited by 773
Fox, N. K., Brenner, S. E., & Chandonia, J.-M. (2014) · Nucleic Acids Research · doi:10.1093/nar/gkt1240 · Cited by 762
Evaluating Protein Transfer Learning with TAPE
Rao, R., Bhattacharya, N., Thomas, N., Duan, Y., Chen, X., Canny, J., Abbeel, P., & Song, Y. S. (2019) · doi:10.1101/676825 · Cited by 719 · cited 8×
“On protein-related downstream benchmarks (Rao et al., 2019), CPCProt achieves comparable results, despite using 0.”
BackgroundMethodResultInfluentialModeling the language of life – Deep Learning Protein Sequences
Heinzinger, M., Elnaggar, A., Wang, Y., Dallago, C., Nechaev, D., Matthes, F., & Rost, B. (2019) · doi:10.1101/614313 · Cited by 606
Local fitness landscape of the green fluorescent protein
Sarkisyan, K. S., Bolotin, D. A., Meer, M. V., Usmanova, D. R., Mishin, A. S., Sharonov, G. V., Ivankov, D. N., Bozhanova, N. G., Baranov, M. S., Soylemez, O., et al. (2016) · Nature · doi:10.1038/nature17995 · Cited by 591
Global analysis of protein folding using massively parallel design, synthesis, and testing
Rocklin, G. J., Chidyausiku, T. M., Goreshnik, I., Ford, A., Houliston, S., Lemak, A., Carter, L., Ravichandran, R., Mulligan, V. K., Chevalier, A., et al. (2017) · Science · doi:10.1126/science.aan0693 · Cited by 501 · cited 2×
“Given the diversity of biological use cases, embedding evaluations should be made on a case-by-case basis.”
BackgroundMethodOn Variational Bounds of Mutual Information
Poole, B., Ozair, S., Oord, A. V. D., Alemi, A. A., & Tucker, G. (2019) · arXiv · doi:10.48550/arxiv.1905.06922 · Cited by 480 · cited 2×
“Theoretically, the InfoNCE estimator requires a large number of negative samples to reduce variance (Poole et al., 2019). Further, as formalized in Wang & Isola (2020), in the limit of infinite negative samples, the InfoMax objective has the desirable property of directly optimizing for the properties of alignment (si…”
BackgroundTian, Y., Krishnan, D., & Isola, P. (2019) · arXiv · doi:10.48550/arxiv.1906.05849 · Cited by 465
Korber, B. T., Farber, R. M., Wolpert, D. H., & Lapedes, A. S. (1993) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.90.15.7176 · Cited by 257
Learning protein sequence embeddings using information from structure
Bepler, T., & Berger, B. (2019) · arXiv · doi:10.48550/arxiv.1902.08661 · Cited by 219 · cited 2×
“On protein-related downstream benchmarks [29], CPCProt achieves comparable results, despite using 0.9% the number of pretraining parameters of the largest model [3] and 8.9% of the number of parameters of the smallest neural network model [6], as illustrated in Figure 1.”
BackgroundMethodInfluentialUnsupervised Pretraining Transfers Well Across Languages
Riviere, M., Joulin, A., Mazare, P.-E., & Dupoux, E. (2020) · ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp40776.2020.9054548 · Cited by 186
Cited by
Language models enable zero-shot prediction of the effects of mutations on protein function
Meier, J., Rao, R., Verkuil, R., Liu, J., Sercu, T., & Rives, A. (2021) · doi:10.1101/2021.07.09.450648 · Cited by 635
Self-supervised learning in medicine and healthcare
Krishnan, R., Rajpurkar, P., & Topol, E. J. (2022) · Nature Biomedical Engineering · doi:10.1038/s41551-022-00914-1 · Cited by 605
“In addition to using generative techniques, a self-supervised contrastive-learning task can be constructed by creating positive protein fragments taken from the same protein sequence, as well as negative pairs from fragments from different sequences 43 .”
MethodRao, R., Liu, J., Verkuil, R., Meier, J., Canny, J. F., Abbeel, P., Sercu, T., & Rives, A. (2021) · doi:10.1101/2021.02.12.430858 · Cited by 350
Convolutions are competitive with transformers for protein sequence pretraining
Yang, K. K., Fusi, N., & Lu, A. X. (2024) · Cell Systems · doi:10.1016/j.cels.2024.01.008 · Cited by 137
Self-citationOrthrus: toward evolutionary and functional RNA foundation models
Fradkin, P., Shi, R. “., Dalal, T., Isaev, K., Frey, B. J., Lee, L. J., Morris, Q., & Wang, B. (2026) · Nature Methods · doi:10.1038/s41592-026-03064-3 · Cited by 22
Metadata
- Authors
- Lu, A. X., Zhang, H., Ghassemi, M., & Moses, A.
- Venue
- bioRxiv (Cold Spring Harbor Laboratory)
- Publisher
- openRxiv
- Published
- 2020-09-06
- OpenAlex
- W3083386021
- Cited by
- 70
- References
- 25
- Type
- posted content
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.