64: Pisces: Multi-modal augmentation for drug combination prediction

Xu H et al., Cell Genomics - This episode covers Pisces, a machine-learning framework that augments sparse drug-pair datasets by creating multiple modality-based views per drug to improve prediction of drug synergy, xenograft responses, and drug-drug interactions. Key terms: drug combination, data augmentation, multimodal, machine learning, drug-drug interaction.

Study Highlights:
The authors introduce Pisces, which embeds eight drug modalities and combines them pairwise to expand each drug-pair into up to 64 training views, allowing the model to handle missing modalities. Pisces applies contrastive alignment, noisy-label top-k aggregation, and ResNet integration to produce robust predictions. The method attains state-of-the-art performance on cell-line (GDSC-combo), xenograft tumor response, and drug–drug interaction benchmarks. Interpreting Pisces predictions with a genetic interaction network revealed a BRCA drug-sensitive pathway validated in TCGA with survival differences.

Conclusion:
Multi-modal data augmentation by treating modality pairs as independent views enables Pisces to substantially improve prediction of drug synergy, xenograft tumor response, and DDIs and to generate biologically and clinically relevant hypotheses such as a BRCA drug-sensitive pathway.

Music:
Enjoy the music based on this article at the end of the episode.

Article title:
Pisces: A multi-modal data augmentation approach for drug combination synergy prediction

First author:
Xu H

Journal:
Cell Genomics

DOI:
10.1016/j.xgen.2025.100892

Reference:
Xu H., Lin J., Woicik A., Liu Z., Ma J., Zhang S., Poon H., Wang L., Wang S. Pisces: A multi-modal data augmentation approach for drug combination synergy prediction. Cell Genomics. 2025;5:100892. https://doi.org/10.1016/j.xgen.2025.100892

License:
This episode is based on an open-access article published under the Creative Commons Attribution 4.0 International License (CC BY 4.0) – https://creativecommons.org/licenses/by/4.0/

Support:
Base by Base is independent and ad-free — no sponsors, no paywall. If an episode was worth your time, chip in and keep the papers audited and the original songs coming:
❤️ Support monthly: https://buy.stripe.com/cNifZhclVebvagk2JDgEg01
☕ One-time donation: https://donate.stripe.com/7sY4gz71B2sN3RWac5gEg00
More at basebybase.com

On PaperCast Base by Base you'll discover the latest in genomics, functional genomics, structural genomics, and proteomics.

Episode link: https://basebybase.com/episodes/base-by-base-64-garbled-pdf

QC:
This episode was checked against the original article PDF and publication metadata for the episode release published on 2025-07-03.

QC Scope:
- article metadata and core scientific claims from the narration
- excludes analogies, intro/outro, and music
- transcript coverage: Audited transcript sections describing Pisces architecture, data augmentation (64 views), the noisy-aggregation approach, key experimental results (cell lines, unseen drug pairs, xenografts, DDI), BRCA pathway/TCGA validation, and stated limitations.
- transcript topics: Problem of drug synergy prediction and data scarcity; Pisces multimodal data augmentation concept; Eight modalities per drug and 64 augmented views; Projector, augmentor, and aggregator architecture; Noisy label learning and top-k aggregation; Evaluation on GDSC cell lines (unseen drug combos, unseen cell lines)

QC Summary:
- factual score: 10/10
- metadata score: 10/10
- supported core claims: 7
- claims flagged for review: 0
- metadata checks passed: 4
- metadata issues found: 0

Metadata Audited:
- article_doi
- article_title
- article_journal
- license

Factual Items Audited:
- Pisces expands data by 64× by using eight modalities per drug...