The Power of Universal Contextualized Protein Embeddings in Cross-species Protein Function Prediction
Computationally annotating proteins with a molecular function is a difficult problem that is made even harder due to the limited amount of available labeled protein training data. Unsupervised protein embeddings partly circumvent this limitation by learning a universal protein representation from ma...
Guardado en:
Autores principales: | , , |
---|---|
Formato: | article |
Lenguaje: | EN |
Publicado: |
SAGE Publishing
2021
|
Materias: | |
Acceso en línea: | https://doaj.org/article/6517605b2ea149948d0f7ac12058a55c |
Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
id |
oai:doaj.org-article:6517605b2ea149948d0f7ac12058a55c |
---|---|
record_format |
dspace |
spelling |
oai:doaj.org-article:6517605b2ea149948d0f7ac12058a55c2021-12-03T23:03:40ZThe Power of Universal Contextualized Protein Embeddings in Cross-species Protein Function Prediction1176-934310.1177/11769343211062608https://doaj.org/article/6517605b2ea149948d0f7ac12058a55c2021-12-01T00:00:00Zhttps://doi.org/10.1177/11769343211062608https://doaj.org/toc/1176-9343Computationally annotating proteins with a molecular function is a difficult problem that is made even harder due to the limited amount of available labeled protein training data. Unsupervised protein embeddings partly circumvent this limitation by learning a universal protein representation from many unlabeled sequences. Such embeddings incorporate contextual information of amino acids, thereby modeling the underlying principles of protein sequences insensitive to the context of species. We used an existing pre-trained protein embedding method and subjected its molecular function prediction performance to detailed characterization, first to advance the understanding of protein language models, and second to determine areas of improvement. Then, we applied the model in a transfer learning task by training a function predictor based on the embeddings of annotated protein sequences of one training species and making predictions on the proteins of several test species with varying evolutionary distance. We show that this approach successfully generalizes knowledge about protein function from one eukaryotic species to various other species, outperforming both an alignment-based and a supervised-learning-based baseline. This implies that such a method could be effective for molecular function prediction in inadequately annotated species from understudied taxonomic kingdoms.Irene van den BentStavros MakrodimitrisMarcel ReindersSAGE PublishingarticleEvolutionQH359-425ENEvolutionary Bioinformatics, Vol 17 (2021) |
institution |
DOAJ |
collection |
DOAJ |
language |
EN |
topic |
Evolution QH359-425 |
spellingShingle |
Evolution QH359-425 Irene van den Bent Stavros Makrodimitris Marcel Reinders The Power of Universal Contextualized Protein Embeddings in Cross-species Protein Function Prediction |
description |
Computationally annotating proteins with a molecular function is a difficult problem that is made even harder due to the limited amount of available labeled protein training data. Unsupervised protein embeddings partly circumvent this limitation by learning a universal protein representation from many unlabeled sequences. Such embeddings incorporate contextual information of amino acids, thereby modeling the underlying principles of protein sequences insensitive to the context of species. We used an existing pre-trained protein embedding method and subjected its molecular function prediction performance to detailed characterization, first to advance the understanding of protein language models, and second to determine areas of improvement. Then, we applied the model in a transfer learning task by training a function predictor based on the embeddings of annotated protein sequences of one training species and making predictions on the proteins of several test species with varying evolutionary distance. We show that this approach successfully generalizes knowledge about protein function from one eukaryotic species to various other species, outperforming both an alignment-based and a supervised-learning-based baseline. This implies that such a method could be effective for molecular function prediction in inadequately annotated species from understudied taxonomic kingdoms. |
format |
article |
author |
Irene van den Bent Stavros Makrodimitris Marcel Reinders |
author_facet |
Irene van den Bent Stavros Makrodimitris Marcel Reinders |
author_sort |
Irene van den Bent |
title |
The Power of Universal Contextualized Protein Embeddings in Cross-species Protein Function Prediction |
title_short |
The Power of Universal Contextualized Protein Embeddings in Cross-species Protein Function Prediction |
title_full |
The Power of Universal Contextualized Protein Embeddings in Cross-species Protein Function Prediction |
title_fullStr |
The Power of Universal Contextualized Protein Embeddings in Cross-species Protein Function Prediction |
title_full_unstemmed |
The Power of Universal Contextualized Protein Embeddings in Cross-species Protein Function Prediction |
title_sort |
power of universal contextualized protein embeddings in cross-species protein function prediction |
publisher |
SAGE Publishing |
publishDate |
2021 |
url |
https://doaj.org/article/6517605b2ea149948d0f7ac12058a55c |
work_keys_str_mv |
AT irenevandenbent thepowerofuniversalcontextualizedproteinembeddingsincrossspeciesproteinfunctionprediction AT stavrosmakrodimitris thepowerofuniversalcontextualizedproteinembeddingsincrossspeciesproteinfunctionprediction AT marcelreinders thepowerofuniversalcontextualizedproteinembeddingsincrossspeciesproteinfunctionprediction AT irenevandenbent powerofuniversalcontextualizedproteinembeddingsincrossspeciesproteinfunctionprediction AT stavrosmakrodimitris powerofuniversalcontextualizedproteinembeddingsincrossspeciesproteinfunctionprediction AT marcelreinders powerofuniversalcontextualizedproteinembeddingsincrossspeciesproteinfunctionprediction |
_version_ |
1718373081955696640 |