An auditable framework for CSF-associated receptor evidence: selective annotation, contamination controls and transfer limits

Single-cell receptor maps can connect cerebrospinal fluid (CSF) proteins to plausible trigeminal targets, but their interpretation depends on biological replication, cell identity, component co-detection and technical background. We developed an auditable framework that preserves these evidence levels and compares proposed computational improvements against matched baselines. The project encompasses 251,826 retained real observations in 60 technical libraries, including 117,405 original trigeminal/extension observations, 22,590 adjacent dorsal root ganglion (DRG) observations and an external author-label benchmark of 111,831 human DRG nuclei from seven donors. A frozen panel of 28 genes and eight component pairs retains unavailable measurements and fixed denominators. Whole-donor/pool holdout and exclusion of candidate features improved internal agreement for several classes: mouse Schwann precision/recall increased from 86.2%/75.1% with fixed markers to 98.4%/95.8% with reference mapping. External transfer exposed substantial limitations. Class-specific thresholds increased acceptance from 90.9% to 93.4%, while agreement among accepted nuclei changed from 82.4% to 82.1%; Schwann precision remained 22.9%. Strict consensus achieved 95.3% agreement at 65.4% acceptance and rejected 99.4% of 659 out-of-reference adipocytes. Across known-count simulations, conservative correction reduced mean zero-contamination native-count loss from 8.94% to 4.12%, including both correct and corrupted partitions, but worsened reconstruction relative to default correction at high contamination. Uncorrected counts had the lowest average reconstruction error across the specified grid. Real-data remedy comparisons changed no receptor pair-state classifications. All 36 annotation and 64 synthetic correction fits converged, and 1,164,055 numerical checks passed. These checks establish computational consistency, not experimental identity or receptor function.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-19
DOI
https://doi.org/10.5281/zenodo.24369229
Primary Topic
Single-cell and spatial transcriptomics
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

An auditable framework for CSF-associated receptor evidence: selective annotation, contamination controls and transfer limits

S. M. Abdulla Al Mamun
Zenodo (CERN European Organization for Nuclear Research)
Single-cell and spatial transcriptomics
preprint

An auditable framework for CSF-associated receptor evidence: selective annotation, contamination controls and transfer limits

S. M. Abdulla Al Mamun
preprint en

Abstract

Single-cell receptor maps can connect cerebrospinal fluid (CSF) proteins to plausible trigeminal targets, but their interpretation depends on biological replication, cell identity, component co-detection and technical background. We developed an auditable framework that preserves these evidence levels and compares proposed computational improvements against matched baselines. The project encompasses 251,826 retained real observations in 60 technical libraries, including 117,405 original trigeminal/extension observations, 22,590 adjacent dorsal root ganglion (DRG) observations and an external author-label benchmark of 111,831 human DRG nuclei from seven donors. A frozen panel of 28 genes and eight component pairs retains unavailable measurements and fixed denominators. Whole-donor/pool holdout and exclusion of candidate features improved internal agreement for several classes: mouse Schwann precision/recall increased from 86.2%/75.1% with fixed markers to 98.4%/95.8% with reference mapping. External transfer exposed substantial limitations. Class-specific thresholds increased acceptance from 90.9% to 93.4%, while agreement among accepted nuclei changed from 82.4% to 82.1%; Schwann precision remained 22.9%. Strict consensus achieved 95.3% agreement at 65.4% acceptance and rejected 99.4% of 659 out-of-reference adipocytes. Across known-count simulations, conservative correction reduced mean zero-contamination native-count loss from 8.94% to 4.12%, including both correct and corrupted partitions, but worsened reconstruction relative to default correction at high contamination. Uncorrected counts had the lowest average reconstruction error across the specified grid. Real-data remedy comparisons changed no receptor pair-state classifications. All 36 annotation and 64 synthetic correction fits converged, and 1,164,055 numerical checks passed. These checks establish computational consistency, not experimental identity or receptor function.

Zenodo (CERN European Organization for Nuclear Research)
University of Manchester (GB)
Single-cell and spatial transcriptomics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon