PortPred: exploiting deep learning embeddings of amino acid sequences for the identification of transporter proteins and their substrates

Abstract

The physiology of every living cell is regulated at some level by transporter proteins which constitute a relevant portion of membrane-bound proteins and are involved in the movement of ions, small and macromolecules across bio-membranes. The importance of transporter proteins is unquestionable. The prediction and study of previously unknown transporters can lead to the discovery of new biological pathways, drugs and treatments. Here we present PortPred, a tool to accurately identify transporter proteins and their substrate starting from the protein amino acid sequence. PortPred successfully combines pre-trained deep learning-based protein embeddings and machine learning classification approaches and outperforms other state-of-the-art methods. In addition, we present a comparison of the most promising protein sequence embeddings (Unirep, SeqVec, ProteinBERT, ESM-1b) and their performances for this specific task.

Semantic search using protein large language models detects class II microcins in bacterial genomes

“…At the core of these models are so-called embeddings, which encode protein characteristics into a high-dimensional numerical space suitable for many different downstream applications. In particular, recent works have shown that embeddings can be used for the identification of protein homologs, even in cases of low sequence similarity [34; 35; 36].…”

P-PPI: accurate prediction of peroxisomal protein-protein interactions (P-PPI) using deep learning-based protein sequence embeddings

“…The application of deep learning (DL) approaches to encode protein sequences has shown promising results in several tasks such as subcellular [8] and sub-organelle [9][10][11] classification, protein structure [12,13] and function prediction [14][15][16][17], and proteinprotein interactions (PPI) prediction [18,19].…”

“…DL-based embedding predictors offer several advantages. Firstly, they do not require lengthy pre-processing steps, as only the FASTA sequence is necessary for input [9,10,17]. This simplifies the prediction process and saves time.…”

“…This simplifies the prediction process and saves time. Additionally, DL-based embeddings are pre-computed, making the overall procedure computationally efficient [9,10,17]. Lastly, these predictors consistently yield highly accurate results [9,10,17].…”

“…Additionally, DL-based embeddings are pre-computed, making the overall procedure computationally efficient [9,10,17]. Lastly, these predictors consistently yield highly accurate results [9,10,17]. Among the established PPI predictors that incorporate various types of information beyond just the sequence, notable examples include: PIPE4 [21].…”