Journal
Elife
Publication Date
2021
Volume
10
First Page
e70576
Document Type
Open Access Publication
DOI
10.7554/eLife.70576
Rights and Permissions
eLife 2021;10:e70576 DOI: 10.7554/eLife.70576. © 2021, Griffith and Holehouse. This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Recommended Citation
Griffith, Daniel and Holehouse, Alex S., "PARROT is a flexible recurrent neural network framework for analysis of large protein datasets." Elife. 10, e70576 (2021).
https://digitalcommons.wustl.edu/open_access_pubs/10870
elife-70576-supp1-v1.docx (17 kB)
Supplementary file 1 Complete table of performance metrics for phosphosite predictions on the Phospho.ELM (P.ELM) dataset. Standard error, whenever possible, is reported in parentheses.
elife-70576-supp2-v1.docx (18 kB)
Supplementary file 2 Complete table of performance metrics for phosphosite predictions on the PhosPhAt (PPA) datasets. Standard error, whenever possible, is reported in parentheses.
elife-70576-supp3-v1.docx (15 kB)
Supplementary file 3 Average PARROT network training times on different sizes of datasets and with variable hyperparameters. Datasets were created by assigning random values in [–5, 5] to randomly generated protein sequences ~25–35 residues in length. Networks were trained using a NVIDIA TU116 GPU. Three replicate PARROT networks were trained for each specified set of hyperparameters and dataset.
elife-70576-transrepform1-v1.docx (111 kB)
Transparent reporting form
Figures and figure supplements.pdf (2008 kB)
Supplementary file 1 Complete table of performance metrics for phosphosite predictions on the Phospho.ELM (P.ELM) dataset. Standard error, whenever possible, is reported in parentheses.
elife-70576-supp2-v1.docx (18 kB)
Supplementary file 2 Complete table of performance metrics for phosphosite predictions on the PhosPhAt (PPA) datasets. Standard error, whenever possible, is reported in parentheses.
elife-70576-supp3-v1.docx (15 kB)
Supplementary file 3 Average PARROT network training times on different sizes of datasets and with variable hyperparameters. Datasets were created by assigning random values in [–5, 5] to randomly generated protein sequences ~25–35 residues in length. Networks were trained using a NVIDIA TU116 GPU. Three replicate PARROT networks were trained for each specified set of hyperparameters and dataset.
elife-70576-transrepform1-v1.docx (111 kB)
Transparent reporting form
Figures and figure supplements.pdf (2008 kB)
