Details
Originalsprache | Englisch |
---|---|
Aufsatznummer | 28 |
Fachzeitschrift | Algorithms |
Jahrgang | 14 |
Ausgabenummer | 1 |
Publikationsstatus | Veröffentlicht - 19 Jan. 2021 |
Abstract
Predicting biological properties of unseen proteins is shown to be improved by the use of protein sequence embeddings. However, these sequence embeddings have the caveat that biological metadata do not exist for each amino acid, in order to measure the quality of each unique learned embedding vector separately. Therefore, current sequence embedding cannot be intrinsically evaluated on the degree of their captured biological information in a quantitative manner. We address this drawback by our approach, dom2vec, by learning vector representation for protein domains and not for each amino acid base, as biological metadata do exist for each domain separately. To perform a reliable quantitative intrinsic evaluation in terms of biology knowledge, we selected the metadata related to the most distinctive biological characteristics of a domain, which are its structure, enzymatic, and molecular function. Notably, dom2vec obtains an adequate level of performance in the intrinsic assessment—therefore, we can draw an analogy between the local linguistic features in natural languages and the domain structure and function information in domain architectures. Moreover, we demonstrate the dom2vec applicability on protein prediction tasks, by comparing it with state-of-the-art sequence embeddings in three downstream tasks. We show that dom2vec outperforms sequence embeddings for toxin and enzymatic function prediction and is comparable with sequence embeddings in cellular location prediction.
ASJC Scopus Sachgebiete
- Mathematik (insg.)
- Theoretische Informatik
- Mathematik (insg.)
- Numerische Mathematik
- Informatik (insg.)
- Theoretische Informatik und Mathematik
- Mathematik (insg.)
- Computational Mathematics
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
in: Algorithms, Jahrgang 14, Nr. 1, 28, 19.01.2021.
Publikation: Beitrag in Fachzeitschrift › Artikel › Forschung › Peer-Review
}
TY - JOUR
T1 - Capturing protein domain structure and function using self-supervision on domain architectures
AU - Melidis, Damianos P.
AU - Nejdl, Wolfgang
N1 - Funding Information: Funding: This study was funded by the Ministry for Science and Culture of Lower Saxony Germany (MWK: Ministerium für Wissenschaft und Kultur) within the project “Understanding Cochlear Implant Outcome Variability using Big Data and Machine Learning Approaches”, project id: ZN3429.
PY - 2021/1/19
Y1 - 2021/1/19
N2 - Predicting biological properties of unseen proteins is shown to be improved by the use of protein sequence embeddings. However, these sequence embeddings have the caveat that biological metadata do not exist for each amino acid, in order to measure the quality of each unique learned embedding vector separately. Therefore, current sequence embedding cannot be intrinsically evaluated on the degree of their captured biological information in a quantitative manner. We address this drawback by our approach, dom2vec, by learning vector representation for protein domains and not for each amino acid base, as biological metadata do exist for each domain separately. To perform a reliable quantitative intrinsic evaluation in terms of biology knowledge, we selected the metadata related to the most distinctive biological characteristics of a domain, which are its structure, enzymatic, and molecular function. Notably, dom2vec obtains an adequate level of performance in the intrinsic assessment—therefore, we can draw an analogy between the local linguistic features in natural languages and the domain structure and function information in domain architectures. Moreover, we demonstrate the dom2vec applicability on protein prediction tasks, by comparing it with state-of-the-art sequence embeddings in three downstream tasks. We show that dom2vec outperforms sequence embeddings for toxin and enzymatic function prediction and is comparable with sequence embeddings in cellular location prediction.
AB - Predicting biological properties of unseen proteins is shown to be improved by the use of protein sequence embeddings. However, these sequence embeddings have the caveat that biological metadata do not exist for each amino acid, in order to measure the quality of each unique learned embedding vector separately. Therefore, current sequence embedding cannot be intrinsically evaluated on the degree of their captured biological information in a quantitative manner. We address this drawback by our approach, dom2vec, by learning vector representation for protein domains and not for each amino acid base, as biological metadata do exist for each domain separately. To perform a reliable quantitative intrinsic evaluation in terms of biology knowledge, we selected the metadata related to the most distinctive biological characteristics of a domain, which are its structure, enzymatic, and molecular function. Notably, dom2vec obtains an adequate level of performance in the intrinsic assessment—therefore, we can draw an analogy between the local linguistic features in natural languages and the domain structure and function information in domain architectures. Moreover, we demonstrate the dom2vec applicability on protein prediction tasks, by comparing it with state-of-the-art sequence embeddings in three downstream tasks. We show that dom2vec outperforms sequence embeddings for toxin and enzymatic function prediction and is comparable with sequence embeddings in cellular location prediction.
KW - Enzymatic commission class
KW - Protein domain architectures
KW - Quantitative quality assessment
KW - SCOPe secondary structure class
KW - Word embeddings
UR - http://www.scopus.com/inward/record.url?scp=85099789523&partnerID=8YFLogxK
U2 - 10.3390/a14010028
DO - 10.3390/a14010028
M3 - Article
AN - SCOPUS:85099789523
VL - 14
JO - Algorithms
JF - Algorithms
IS - 1
M1 - 28
ER -