Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.19509902 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901964168626176 |
|---|---|
| author | Song, Chaolin |
| author_facet | Song, Chaolin |
| contents | <p>Table S1 presents the comparative results of MMFGO against several existing methods on the CAFA4 dataset, showing that the proposed model achieves strong performance across multiple evaluation metrics for the three GO categories, namely BP, MF, and CC. Table S2 reports the ablation results for different modalities, illustrating the respective contributions of sequence, text, and structural information to the overall model performance. Table S3 further compares MHIF with several other fusion modules, highlighting the advantage of MHIF in sequence multiview fusion. Figures S1–S3 mainly provide heatmap visualizations of Fmax in cross-species functional annotation tasks. By comparing baseline encoding methods with the multiview fusion strategy enhanced by MHIF, these figures demonstrate the performance improvements from the perspectives of different species, GO branches, and modality representations. Overall, this supplementary file provides more comprehensive comparative, ablation, and visualization evidence to further support the effectiveness of MMFGO in multimodal and multiview protein function prediction.</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_19509902 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | MMFGO_supporting_information Song, Chaolin <p>Table S1 presents the comparative results of MMFGO against several existing methods on the CAFA4 dataset, showing that the proposed model achieves strong performance across multiple evaluation metrics for the three GO categories, namely BP, MF, and CC. Table S2 reports the ablation results for different modalities, illustrating the respective contributions of sequence, text, and structural information to the overall model performance. Table S3 further compares MHIF with several other fusion modules, highlighting the advantage of MHIF in sequence multiview fusion. Figures S1–S3 mainly provide heatmap visualizations of Fmax in cross-species functional annotation tasks. By comparing baseline encoding methods with the multiview fusion strategy enhanced by MHIF, these figures demonstrate the performance improvements from the perspectives of different species, GO branches, and modality representations. Overall, this supplementary file provides more comprehensive comparative, ablation, and visualization evidence to further support the effectiveness of MMFGO in multimodal and multiview protein function prediction.</p> |
| title | MMFGO_supporting_information |
| url | https://doi.org/10.5281/zenodo.19509902 |