64: A Garbled PDF

Xu H et al., Cell Genomics - This episode examines a heavily corrupted PDF provided as the source. The text is dominated by recurring, unreadable tokens (e.g., Wt�mo�, m{yltzk�t{z, k�ryoz�k�t{z) and fragmented sections, preventing clear extraction of aims or results. We walk listeners through what can and cannot be recovered from the file. Key terms: Wt�mo�, m{yltzk�t{z, k�ryoz�k�t{z, oqqom�t�owÞ, J~�rGkzv.

Study Highlights:
The supplied PDF is extensively corrupted and repeatedly contains tokens such as "Wt�mo�", "m{yltzk�t{z", and "k�ryoz�k�t{z" that recur throughout. Sections also reference forms like "oqqom�t�owÞ" and labels such as "J~�rGkzv", suggesting structured headings or entities but unreadable encoding. Because of pervasive formatting and encoding errors the study's aims, methods and results cannot be reliably extracted from the text.

Conclusion:
The PDF text is too corrupted to recover definitive conclusions; a clean source is required for meaningful interpretation.

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., et al.. Pisces: A multi-modal data augmentation approach for drug combination synergy prediction. Cell Genomics, 5, 100892. (2025). 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 – Stripe donations: https://donate.stripe.com/7sY4gz71B2sN3RWac5gEg00

Official website https://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: Substantively audited sections describing Pisces architecture, data augmentation, the 64-view augmenter, the noisy-label aggregator, and the key experimental results (cell lines, unseen drug pairs, 3-drug synergy, in vivo) plus limitations and clinical implications.
- transcript topics: Problem of drug synergy and data scarcity; Multimodal data augmentation concept (Pisces); Eight modalities per drug and universal embedding; The augmenter: 8 x 8 views = 64 augmented views; Noisy label aggregator selecting top 8 predictions; Evaluation on GDSC data and unseen drug pair/cell line splits

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 uses 8 modalities per drug and forms 64 augmented views for each drug pair
- Projector translates cross-modality representations into a shared embedding space
- Aggregator employs noisy label learning and retains the top 8 predictions
- Unseen 2-drug combinations: F1 improves by ~24% over the next-best approach
- Unseen cell lines: F1 improvement > ~10%
- Triplet (3-drug) synergy evaluation: AUROC = 0.8525

QC result: Pass.