TY - GEN
T1 - Spatial Transcriptomics Analysis of Zero-Shot Gene Expression Prediction
AU - Yang, Yan
AU - Hossain, Md Zakir
AU - Li, Xuesong
AU - Rahman, Shafin
AU - Stone, Eric
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
PY - 2024
Y1 - 2024
N2 - Spatial transcriptomics (ST) captures gene expression fine-grained distinct regions (i.e., windows) of a tissue slide. Traditional supervised learning frameworks applied to model ST are constrained to predicting expression of gene types seen during training from slide image windows, failing to generalize to unseen gene types. To overcome this limitation, we propose a semantic guided network, a pioneering zero-shot gene expression prediction framework. Considering a gene type can be described by functionality and phenotype, we dynamically embed a gene type to a vector per its functionality and phenotype, and employ this vector to project slide image windows to gene expression in feature space, unleashing zero-shot expression prediction for unseen gene types. The gene type functionality and phenotype are queried with a carefully designed prompt from a pre-trained large language model. On standard benchmark datasets, we demonstrate competitive zero-shot performance compared to past state-of-the-art supervised learning approaches. Our code is available at https://github.com/Yan98/SGN.
AB - Spatial transcriptomics (ST) captures gene expression fine-grained distinct regions (i.e., windows) of a tissue slide. Traditional supervised learning frameworks applied to model ST are constrained to predicting expression of gene types seen during training from slide image windows, failing to generalize to unseen gene types. To overcome this limitation, we propose a semantic guided network, a pioneering zero-shot gene expression prediction framework. Considering a gene type can be described by functionality and phenotype, we dynamically embed a gene type to a vector per its functionality and phenotype, and employ this vector to project slide image windows to gene expression in feature space, unleashing zero-shot expression prediction for unseen gene types. The gene type functionality and phenotype are queried with a carefully designed prompt from a pre-trained large language model. On standard benchmark datasets, we demonstrate competitive zero-shot performance compared to past state-of-the-art supervised learning approaches. Our code is available at https://github.com/Yan98/SGN.
KW - Computational pathology
KW - Gene expression prediction
KW - Spatial transcriptomics
KW - Tissue slide image
KW - Zero-shot learning
UR - http://www.scopus.com/inward/record.url?scp=85207663180&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-72083-3_46
DO - 10.1007/978-3-031-72083-3_46
M3 - Conference contribution
AN - SCOPUS:85207663180
SN - 9783031720826
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 492
EP - 502
BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2024 - 27th International Conference, Proceedings
A2 - Linguraru, Marius George
A2 - Dou, Qi
A2 - Feragen, Aasa
A2 - Giannarou, Stamatia
A2 - Glocker, Ben
A2 - Lekadir, Karim
A2 - Schnabel, Julia A.
PB - Springer Science and Business Media Deutschland GmbH
T2 - 27th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2024
Y2 - 6 October 2024 through 10 October 2024
ER -