Knowledge-based inductive bias and domain adaptation for cell type annotation

Abstract Measurement techniques often result in domain gaps among batches of cellular data from a specific modality. The effectiveness of cross-batch annotation methods is influenced by inductive bias, which refers to a set of assumptions that describe the behavior of model predictions. Different an...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhenchao Tang, Guanxing Chen, Shouzhi Chen, Haohuai He, Linlin You, Calvin Yu-Chian Chen
Format: Article
Language:English
Published: Nature Portfolio 2024-11-01
Series:Communications Biology
Online Access:https://doi.org/10.1038/s42003-024-07171-9
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850194811981660160
author Zhenchao Tang
Guanxing Chen
Shouzhi Chen
Haohuai He
Linlin You
Calvin Yu-Chian Chen
author_facet Zhenchao Tang
Guanxing Chen
Shouzhi Chen
Haohuai He
Linlin You
Calvin Yu-Chian Chen
author_sort Zhenchao Tang
collection DOAJ
description Abstract Measurement techniques often result in domain gaps among batches of cellular data from a specific modality. The effectiveness of cross-batch annotation methods is influenced by inductive bias, which refers to a set of assumptions that describe the behavior of model predictions. Different annotation methods possess distinct inductive biases, leading to varying degrees of generalizability and interpretability. Given that certain cell types exhibit unique functional patterns, we hypothesize that the inductive biases of cell annotation methods should align with these biological patterns to produce meaningful predictions. In this study, we propose KIDA, Knowledge-based Inductive bias and Domain Adaptation. The knowledge-based inductive bias constrains the prediction rules learned from the reference dataset, composed of multiple batches, to functional patterns relevant to biology, thereby enhancing the generalization of the model to unseen batches. Since the query dataset also contains gaps from multiple batches, KIDA’s domain adaptation employs pseudo labels for self-knowledge distillation, effectively narrowing the distribution gap between model predictions and the query dataset. Benchmark experiments demonstrate that KIDA is capable of achieving accurate cross-batch cell type annotation.
format Article
id doaj-art-9896a52e9c4c4caf9ae7fc54eecfb257
institution OA Journals
issn 2399-3642
language English
publishDate 2024-11-01
publisher Nature Portfolio
record_format Article
series Communications Biology
spelling doaj-art-9896a52e9c4c4caf9ae7fc54eecfb2572025-08-20T02:13:55ZengNature PortfolioCommunications Biology2399-36422024-11-017111010.1038/s42003-024-07171-9Knowledge-based inductive bias and domain adaptation for cell type annotationZhenchao Tang0Guanxing Chen1Shouzhi Chen2Haohuai He3Linlin You4Calvin Yu-Chian Chen5Artificial Intelligence Medical Research Center, School of Intelligent Systems Engineering, Shenzhen Campus of Sun Yat-sen UniversityArtificial Intelligence Medical Research Center, School of Intelligent Systems Engineering, Shenzhen Campus of Sun Yat-sen UniversityArtificial Intelligence Medical Research Center, School of Intelligent Systems Engineering, Shenzhen Campus of Sun Yat-sen UniversityArtificial Intelligence Medical Research Center, School of Intelligent Systems Engineering, Shenzhen Campus of Sun Yat-sen UniversityArtificial Intelligence Medical Research Center, School of Intelligent Systems Engineering, Shenzhen Campus of Sun Yat-sen UniversityAI for Science (AI4S)-Preferred Program, School of Electronic and Computer Engineering, Peking University Shenzhen Graduate SchoolAbstract Measurement techniques often result in domain gaps among batches of cellular data from a specific modality. The effectiveness of cross-batch annotation methods is influenced by inductive bias, which refers to a set of assumptions that describe the behavior of model predictions. Different annotation methods possess distinct inductive biases, leading to varying degrees of generalizability and interpretability. Given that certain cell types exhibit unique functional patterns, we hypothesize that the inductive biases of cell annotation methods should align with these biological patterns to produce meaningful predictions. In this study, we propose KIDA, Knowledge-based Inductive bias and Domain Adaptation. The knowledge-based inductive bias constrains the prediction rules learned from the reference dataset, composed of multiple batches, to functional patterns relevant to biology, thereby enhancing the generalization of the model to unseen batches. Since the query dataset also contains gaps from multiple batches, KIDA’s domain adaptation employs pseudo labels for self-knowledge distillation, effectively narrowing the distribution gap between model predictions and the query dataset. Benchmark experiments demonstrate that KIDA is capable of achieving accurate cross-batch cell type annotation.https://doi.org/10.1038/s42003-024-07171-9
spellingShingle Zhenchao Tang
Guanxing Chen
Shouzhi Chen
Haohuai He
Linlin You
Calvin Yu-Chian Chen
Knowledge-based inductive bias and domain adaptation for cell type annotation
Communications Biology
title Knowledge-based inductive bias and domain adaptation for cell type annotation
title_full Knowledge-based inductive bias and domain adaptation for cell type annotation
title_fullStr Knowledge-based inductive bias and domain adaptation for cell type annotation
title_full_unstemmed Knowledge-based inductive bias and domain adaptation for cell type annotation
title_short Knowledge-based inductive bias and domain adaptation for cell type annotation
title_sort knowledge based inductive bias and domain adaptation for cell type annotation
url https://doi.org/10.1038/s42003-024-07171-9
work_keys_str_mv AT zhenchaotang knowledgebasedinductivebiasanddomainadaptationforcelltypeannotation
AT guanxingchen knowledgebasedinductivebiasanddomainadaptationforcelltypeannotation
AT shouzhichen knowledgebasedinductivebiasanddomainadaptationforcelltypeannotation
AT haohuaihe knowledgebasedinductivebiasanddomainadaptationforcelltypeannotation
AT linlinyou knowledgebasedinductivebiasanddomainadaptationforcelltypeannotation
AT calvinyuchianchen knowledgebasedinductivebiasanddomainadaptationforcelltypeannotation