OrganelX web server for sub-peroxisomal and sub-mitochondrial protein localization and peroxisomal target signal detection

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

“…Therefore, we emphasize the adaptability of the ESM-1b embedding for fine-tuned prediction tasks. Additionally, these results align with our prior studies on DL-based protein embeddings, as demonstrated in Anteghini et al 22 and Anteghini et al 26 Our research initially focused on a binary prediction task, distinguishing transporter from non-transporter proteins. PortPred exceeded expectations with an accuracy of 94.53%.…”

“…However, the application of deep learning (DL) approaches to encode protein amino acid sequences has shown promising results for several tasks such as subcellular and sub-organelle classification, protein structure and function prediction, and proteinprotein interactions (PPIs). [20][21][22][23][24][25][26] Following our previous works on the use of sequence embeddings for the prediction of the subcellular localiza-the best protein representation and machine learning classifier. PortPred was also tested against the state-ofthe-art transporters predictors [13][14][15][16][17]19 for either binary classification (transporter vs. not-transporter) and multiclass classification related to the transporter substrates namely: cation, anion, electron, lipid, amino acid, protein/messenger RNA (mRNA), sugar, others.…”

“…All these tools exploit the amino acid composition of the protein sequences. However, the application of deep learning (DL) approaches to encode protein amino acid sequences has shown promising results for several tasks such as subcellular and sub‐organelle classification, protein structure and function prediction, and protein‐protein interactions (PPIs) 20–26 …”

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

“…All these tools exploit the amino acid composition of the protein sequences. However, the application of deep learning (DL) approaches to encode protein amino acid sequences has shown promising results for several tasks such as subcellular and sub-organelle classification, protein structure and function prediction, and protein-protein interactions (PPI) [19][20][21][22][23][24][25][26].…”

“…Following our previous works on the use of sequence embeddings for the prediction of the sub-cellular localization of peroxisomal proteins, [21, 26] we apply a similar framework to the development of PortPred, a prediction tool for the accurate identification of transporter proteins and multi-class classification of their transported substrates.…”