IDP-Bert: Predicting Properties of Intrinsically Disordered Proteins Using Large Language Models DOI Creative Commons
Parisa Mollaei,

Danush Sadasivam,

Chakradhar Guntuboina

и другие.

The Journal of Physical Chemistry B, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 25, 2024

Intrinsically disordered Proteins (IDPs) constitute a large and structureless class of proteins with significant functions. The existence IDPs challenges the conventional notion that biological functions rely on their three-dimensional structures. Despite lacking well-defined spatial arrangements, they exhibit diverse functions, influencing cellular processes shedding light disease mechanisms. However, it is expensive to run experiments or simulations characterize this proteins. Consequently, we designed an ML model relies solely amino acid sequences. In study, introduce IDP-Bert model, deep-learning architecture leveraging Transformers Protein Language Models map sequences directly IDP properties. Our demonstrate accurate predictions properties, including Radius Gyration, end-to-end Decorrelation Time, Heat Capacity.

Язык: Английский

PeptideBERT: A Language Model Based on Transformers for Peptide Property Prediction DOI Creative Commons
Chakradhar Guntuboina,

Adrita Das,

Parisa Mollaei

и другие.

The Journal of Physical Chemistry Letters, Год журнала: 2023, Номер 14(46), С. 10427 - 10434

Опубликована: Ноя. 13, 2023

Recent advances in language models have enabled the protein modeling community with a powerful tool that uses transformers to represent sequences as text. This breakthrough enables sequence-to-property prediction for peptides without relying on explicit structural data. Inspired by recent progress field of large models, we present PeptideBERT, model specifically tailored predicting essential peptide properties such hemolysis, solubility, and nonfouling. The PeptideBERT utilizes ProtBERT pretrained transformer 12 attention heads hidden layers. Through fine-tuning three downstream tasks, our is state art (SOTA) which crucial determining peptide's potential induce red blood cells well nonfouling properties. Leveraging primarily shorter data set negative samples predominantly associated insoluble peptides, showcases remarkable performance.

Язык: Английский

Процитировано

31
IDP-Bert: Predicting Properties of Intrinsically Disordered Proteins Using Large Language Models DOI Creative Commons
Parisa Mollaei,

Danush Sadasivam,

Chakradhar Guntuboina

и другие.

The Journal of Physical Chemistry B, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 25, 2024

Intrinsically disordered Proteins (IDPs) constitute a large and structureless class of proteins with significant functions. The existence IDPs challenges the conventional notion that biological functions rely on their three-dimensional structures. Despite lacking well-defined spatial arrangements, they exhibit diverse functions, influencing cellular processes shedding light disease mechanisms. However, it is expensive to run experiments or simulations characterize this proteins. Consequently, we designed an ML model relies solely amino acid sequences. In study, introduce IDP-Bert model, deep-learning architecture leveraging Transformers Protein Language Models map sequences directly IDP properties. Our demonstrate accurate predictions properties, including Radius Gyration, end-to-end Decorrelation Time, Heat Capacity.

Язык: Английский

Процитировано

2