Glycan reachability analysis: A bottleneck-aware framework for inferring tissue-specific glycan biosynthetic potential from transcriptomics
- 1Biomedical and Health Informatics Unit, Department of Integrated Health Sciences, Nagoya University Graduate School of Medicine, Nagoya, Aichi, Japan
Abstract
Glycan biosynthesis requires the coordinated expression of glycosyltransferases, modifying enzymes, and nucleotide-sugar synthesis and transport machinery. Existing computational tools predict glycan structures from gene expression using binary thresholds, losing quantitative information about relative biosynthetic capacity across tissues. Here we present glycan biosynthetic reachability analysis, which integrates expression-based Z-scores across curated pathway steps using AND/OR logic and minimum aggregation to produce continuous, tissue-comparable scores and an explicit expression-limiting step. Applied to 17,382 GTEx v8 RNA-seq samples across 54 human tissue types, reachability resolved quantitative differences hidden by presence/absence calls; for example, pancreas was 96% binary-positive for the sLeX pathway but had low median reachability (Z=−1.86). Bottleneck stability was evaluated across all 19 multi-step metrics. In independent HEK293 knockout glycomics, the minimum score contained information beyond a within-knockout permutation null but did not outperform naive mean aggregation or binary topology. Nested leave-one-knockout-out selection favored a relaxed low quantile (q = 0.2) rather than validating the strict minimum. Within-GTEx associations between reachability and signaling-response transcripts are reported only as transcriptomic coherence because predictors and readouts share the same RNA-seq source. The mouse tissue-glycome comparison remained null. Reachability is therefore a hypothesis-generating rank of transcriptomic potential, not a measure of enzymatic activity or glycan abundance. Author summary: Cells are coated with complex sugars called glycans, which help control how cells interact, respond to signals, and participate in disease. A tissue can make a glycan only when all steps in the required biosynthetic pathway are available. I developed glycan reachability analysis to use gene-expression data to rank this biosynthetic potential. The method treats glycan production like a production line and focuses on the least-expressed required step. Across 54 human tissue types, the resulting score revealed quantitative differences that simple present-or-absent rules missed. For example, the pancreas expressed every gene required for sialyl Lewis X, but many were expressed at low levels. Tests using independent glycan measurements from genetically modified cells showed that the score contained useful information, although it did not outperform simpler alternatives. Associations with signaling-related genes were treated only as patterns within the same gene-expression data and not as evidence that glycans caused signaling changes. The identity of the weakest step was stable in most, but not all, tissue–pathway comparisons, and comparison with a mouse tissue glycome was inconclusive. Reachability should therefore be used to prioritize tissues and pathways for experimental testing, not as a replacement for directly measuring glycans.
Related
- Quantifying polymorphism and divergence from epigenetic data: a framework for inferring the action of selection
- WDGTA: a wavelet-denoised graph attention framework for spatial transcriptomics deconvolution.
- Serval: a modular framework for decoding imaging based spatial transcriptomics data.
- OT-knn: A neighborhood-aware optimal transport framework for aligning spatial transcriptomics data
- Editorial: Biosynthetic pathways and metabolic engineering of bioactive natural products from medicinal plants
BibTeX
@article{matsui2026glycan,
title = {{Glycan reachability analysis: A bottleneck-aware framework for inferring tissue-specific glycan biosynthetic potential from transcriptomics}},
author = {Yusuke Matsui},
journal = {PLOS Computational Biology},
year = {2026},
month = sep,
doi = {10.1371/journal.pcbi.1014655},
url = {https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014655},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Glycan reachability analysis: A bottleneck-aware framework for inferring tissue-specific glycan biosynthetic potential from transcriptomics
AU - Yusuke Matsui
PY - 2026
DA - 2026/09/24
JO - PLOS Computational Biology
DO - 10.1371/journal.pcbi.1014655
UR - https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014655
AB - Glycan biosynthesis requires the coordinated expression of glycosyltransferases, modifying enzymes, and nucleotide-sugar synthesis and transport machinery. Existing computational tools predict glycan structures from gene expression using binary thresholds, losing quantitative information about relative biosynthetic capacity across tissues. Here we present glycan biosynthetic reachability analysis, which integrates expression-based Z-scores across curated pathway steps using AND/OR logic and minimum aggregation to produce continuous, tissue-comparable scores and an explicit expression-limiting step. Applied to 17,382 GTEx v8 RNA-seq samples across 54 human tissue types, reachability resolved quantitative differences hidden by presence/absence calls; for example, pancreas was 96% binary-positive for the sLeX pathway but had low median reachability (Z=−1.86). Bottleneck stability was evaluated across all 19 multi-step metrics. In independent HEK293 knockout glycomics, the minimum score contained information beyond a within-knockout permutation null but did not outperform naive mean aggregation or binary topology. Nested leave-one-knockout-out selection favored a relaxed low quantile (q = 0.2) rather than validating the strict minimum. Within-GTEx associations between reachability and signaling-response transcripts are reported only as transcriptomic coherence because predictors and readouts share the same RNA-seq source. The mouse tissue-glycome comparison remained null. Reachability is therefore a hypothesis-generating rank of transcriptomic potential, not a measure of enzymatic activity or glycan abundance. Author summary: Cells are coated with complex sugars called glycans, which help control how cells interact, respond to signals, and participate in disease. A tissue can make a glycan only when all steps in the required biosynthetic pathway are available. I developed glycan reachability analysis to use gene-expression data to rank this biosynthetic potential. The method treats glycan production like a production line and focuses on the least-expressed required step. Across 54 human tissue types, the resulting score revealed quantitative differences that simple present-or-absent rules missed. For example, the pancreas expressed every gene required for sialyl Lewis X, but many were expressed at low levels. Tests using independent glycan measurements from genetically modified cells showed that the score contained useful information, although it did not outperform simpler alternatives. Associations with signaling-related genes were treated only as patterns within the same gene-expression data and not as evidence that glycans caused signaling changes. The identity of the weakest step was stable in most, but not all, tissue–pathway comparisons, and comparison with a mouse tissue glycome was inconclusive. Reachability should therefore be used to prioritize tissues and pathways for experimental testing, not as a replacement for directly measuring glycans.
SN - 1553-7358
ER -