Modelpublic

mevol/BiomedNLP-PubMedBERT-ProteinStructure-NER-v1.4

sourceHugging Facemitupdated 2y agoView on Hugging Face

0likes36downloads

Model Card

Feature	Description
Name	`BiomedNLP-PubMedBERT-ProteinStructure-NER-v1.4`
Default Pipeline	`transformer`, `ner`
Components	`transformer`, `ner`
Vectors	0 keys, 0 unique vectors (0 dimensions)
Sources	n/a
License	n/a
Author	[Melanie Vollmar]()

Label Scheme

<summary>View label scheme (19 labels for 1 components)</summary>

Component	Labels
`ner`	"chemical", "complexassembly", "evidence", "experimentalmethod", "gene", "mutant", "oligomericstate", "protein", "proteinstate", "proteintype", "ptm", "residuename", "residuenamenumber", "residuenumber", "residuerange", "site", "species", "structureelement", "taxonomydomain"

</details>

Scores for entity types

entity type	precision	recall	F1	sample number
"chemical"	0.90	0.93	0.92	390
"complex_assembly"	0.88	0.91	0.89	162
"evidence"	0.86	0.89	0.88	272
"experimental_method"	0.73	0.76	0.75	240
"gene"	0.89	0.86	0.88	66
"mutant"	0.93	0.95	0.94	495
"oligomeric_state"	0.88	1.00	0.93	64
"protein"	0.97	0.97	0.97	1017
"protein_state"	0.78	0.85	0.81	363
"protein_type"	0.84	0.90	0.87	262
"ptm"	0.64	0.81	0.71	37
"residue_name"	0.97	0.92	0.94	84
"residuenamenumber"	0.98	0.99	0.99	487
"residue_number"	1.00	0.93	0.96	14
"residue_range"	0.86	0.91	0.89	47
"site"	0.83	0.86	0.85	139
"species"	0.97	1.00	0.98	59
"structure_element"	0.91	0.92	0.91	677
"taxonomy_domain"	0.97	0.96	0.97	73

Data and annotations

The dataset can be found here: https://huggingface.co/datasets/mevol/proteinstructureNERmodelv1.4

Citation

Vollmar, M., Tirunagari, S., Harrus, D. et al. Dataset from a human-in-the-loop approach to identify functionally important protein residues from literature. Sci Data 11, 1032 2024 https://doi.org/10.1038/s41597-024-03841-9