rewirebio.iobenchmarks
Configuration

Ladon, pretrained encoder, isoform percentages imputed from gene-level counts

Ladon given only gene-level expression of a sample and asked to impute each isoform's share of its gene's RNA.

6 evaluations · 6 results

Overview

An explanatory profile has not yet been reviewed for this record.

Consult the linked sources for architecture or protocol details. Missing evidence is not evidence of a missing capability.

Model Capabilities & Hardware Requirements

Deployment facts recorded for Ladon, pretrained encoder, isoform percentages imputed from gene-level counts. Fields the record does not carry are marked not recorded.

ParametersNot recorded
Minimum VRAMNot recorded
Inference LatencyNot recorded
Context WindowNot recorded
AccessNot recorded
LicenseNot recorded

Evaluated Checkpoint: Pretrained model; pretraining excluded the TCGA cohorts and included the ENCODE4 and blood bulk reference libraries

Evaluations and results

6 evaluations · 6 results. Different protocols are not a single leaderboard.

Sorted by Accuracy (share of the top 5% of genes (460) by measured isoform change whose predicted direction of change in main-isoform percentage matches the bulk reference; 26 COVID-19 draws against four healthy draws) (higher is better). The best value in each column is highlighted. Decimals are rounded for display; each value links to the printed value and its source.

Tested configurationProtocol and datasetAccuracy (share of the top 5% of genes (460) by measured isoform change whose predicted direction of change in main-isoform percentage matches the bulk reference; 26 COVID-19 draws against four healthy draws)Accuracy (share of the top 5% of genes (460) by measured isoform change whose predicted direction of change in main-isoform percentage matches the bulk reference; mean within-person change day 0 to day 14, nine people)MAE (mean absolute error of INSR isoform percentage, 166 TCGA stomach tumours)Pearson (median over 14,526 gene and cancer-type combinations (1,572 tumours) of Pearson r between predicted and measured isoform percentages, each isoform centred on its mean across tumours)Pearson (Pearson r between predicted and bulk-measured changes in isoform percentage, 1,402 within-donor organ-pair comparisons (837 GTEx-selected contrasts, 287 genes, seven donors))Pearson (Pearson r between predicted and measured INSR isoform percentages, 166 TCGA stomach tumours with INSR at least 10 TPM)
Ladon, pretrained encoder, isoform percentages imputed from gene-level countsEvaluationBlood single-cell to isoform, direction accuracy for COVID-19 against healthy draws, top 5% of genes, Giroux blood draws for isoform direction (35 draws, 17 people)70%Not reportedNot reportedNot reportedNot reportedNot reported
Ladon, pretrained encoder, isoform percentages imputed from gene-level countsEvaluationINSR isoform split in TCGA stomach tumours, mean absolute error, TCGA stomach adenocarcinomas with INSR at least 10 TPM (166)Not reportedNot reported4.9%Not reportedNot reportedNot reported
Ladon, pretrained encoder, isoform percentages imputed from gene-level countsEvaluationINSR isoform split in TCGA stomach tumours, Pearson r, TCGA stomach adenocarcinomas with INSR at least 10 TPM (166)Not reportedNot reportedNot reportedNot reportedNot reported0.93
Ladon, pretrained encoder, isoform percentages imputed from gene-level countsEvaluationIsoform variation in eight held-out TCGA cancer types, median Pearson r over gene and cancer-type pairs, TCGA tumours of eight held-out cancer types for isoform recovery (1,572)Not reportedNot reportedNot reported0.81Not reportedNot reported
Ladon, pretrained encoder, isoform percentages imputed from gene-level countsEvaluationENCODE4 within-donor organ pairs, Pearson r of predicted against measured isoform change, ENCODE4 within-donor organ-pair isoform contrasts (1,402 comparisons)Not reportedNot reportedNot reportedNot reported0.8Not reported
Ladon, pretrained encoder, isoform percentages imputed from gene-level countsEvaluationBlood single-cell to isoform, direction accuracy for day 0 to day 14 change, top 5% of genes, Giroux blood draws for isoform direction (35 draws, 17 people)Not reported66%Not reportedNot reportedNot reportedNot reported
All 6 result rows with coverage, uncertainty and sources
Filter evaluations

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configurationProtocol and datasetFindingEvidence and details
Configuration: Ladon, pretrained encoder, isoform percentages imputed from gene-level countsProtocol: INSR isoform split in TCGA stomach tumours, mean absolute error
Dataset subset: TCGA stomach adenocarcinomas with INSR at least 10 TPM (166)
4.9% MAE (mean absolute error of INSR isoform percentage, 166 TCGA stomach tumours)
percent · lower

Uncertainty: Not reported by the source: The caption says intervals are given, but none is printed for MAE.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Ladon, INSR isoform split in 166 stomach tumours, MAE

ladon-20261115-protocol-insr-stomach-mae

Aggregation: Not reported

Ladon: An RNA Foundation Model for Predicting Patient Outcomes (Zenodo PDF) · Figure 11 (page 24) readout 'MAE, points', 'Ladon'
Configuration: Ladon, pretrained encoder, isoform percentages imputed from gene-level countsProtocol: INSR isoform split in TCGA stomach tumours, Pearson r
Dataset subset: TCGA stomach adenocarcinomas with INSR at least 10 TPM (166)
0.93 Pearson (Pearson r between predicted and measured INSR isoform percentages, 166 TCGA stomach tumours with INSR at least 10 TPM)
unitless · higher

Uncertainty: 95% CI 0.91 to 0.95. percentile bootstrap with predictions held fixed, tumours resampled

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Ladon, INSR isoform split in 166 stomach tumours, Pearson r

ladon-20261115-protocol-insr-stomach-pearson

Aggregation: Not reported

Ladon: An RNA Foundation Model for Predicting Patient Outcomes (Zenodo PDF) · Figure 11 (page 24) readout 'r with measured', 'Ladon'
Configuration: Ladon, pretrained encoder, isoform percentages imputed from gene-level countsProtocol: Isoform variation in eight held-out TCGA cancer types, median Pearson r over gene and cancer-type pairs
Dataset subset: TCGA tumours of eight held-out cancer types for isoform recovery (1,572)
0.81 Pearson (median over 14,526 gene and cancer-type combinations (1,572 tumours) of Pearson r between predicted and measured isoform percentages, each isoform centred on its mean across tumours)
unitless · higher

Uncertainty: Not reported by the source: The caption says the 95% bootstrap intervals are narrower than the symbols; no interval is printed.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Ladon, isoform variation in eight held-out cancer types, median Pearson r

ladon-20261115-protocol-isoform-3000-gene-median-r

Aggregation: Not reported

Ladon: An RNA Foundation Model for Predicting Patient Outcomes (Zenodo PDF) · Figure 12a (page 25), row 'Ladon', point label
Configuration: Ladon, pretrained encoder, isoform percentages imputed from gene-level countsProtocol: ENCODE4 within-donor organ pairs, Pearson r of predicted against measured isoform change
Dataset subset: ENCODE4 within-donor organ-pair isoform contrasts (1,402 comparisons)
0.8 Pearson (Pearson r between predicted and bulk-measured changes in isoform percentage, 1,402 within-donor organ-pair comparisons (837 GTEx-selected contrasts, 287 genes, seven donors))
unitless · higher

Uncertainty: Not reported by the source: Drawn as error bars; no interval is printed.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Ladon, ENCODE4 organ-pair isoform change, Pearson r

ladon-20261115-protocol-encode4-organ-pair-r

Aggregation: Not reported

Ladon: An RNA Foundation Model for Predicting Patient Outcomes (Zenodo PDF) · Figure 13 (page 26), right panel, row 'Ladon'; page 27 prose
Configuration: Ladon, pretrained encoder, isoform percentages imputed from gene-level countsProtocol: Blood single-cell to isoform, direction accuracy for COVID-19 against healthy draws, top 5% of genes
Dataset subset: Giroux blood draws for isoform direction (35 draws, 17 people)
70% Accuracy (share of the top 5% of genes (460) by measured isoform change whose predicted direction of change in main-isoform percentage matches the bulk reference; 26 COVID-19 draws against four healthy draws)
percent · higher

Uncertainty: Not reported by the source: No interval is printed for this value.

Coverage: 460/460 genes scored (Top 5% of 9,207 expressed multi-isoform genes)

Author-reported evaluation · Source checked
Methods, coverage and source

Ladon, blood isoform direction accuracy, covid vs healthy, top 5% of genes

ladon-20261115-protocol-blood-direction-covid-vs-healthy

Aggregation: Not reported

Ladon: An RNA Foundation Model for Predicting Patient Outcomes (Zenodo PDF) · Page 27, paragraph beginning 'Finally, we analyzed 35 blood draws', value for 'Ladon' (between COVID-19 and healthy controls)
Configuration: Ladon, pretrained encoder, isoform percentages imputed from gene-level countsProtocol: Blood single-cell to isoform, direction accuracy for day 0 to day 14 change, top 5% of genes
Dataset subset: Giroux blood draws for isoform direction (35 draws, 17 people)
66% Accuracy (share of the top 5% of genes (460) by measured isoform change whose predicted direction of change in main-isoform percentage matches the bulk reference; mean within-person change day 0 to day 14, nine people)
percent · higher

Uncertainty: Not reported by the source: No interval is printed for this value.

Coverage: 460/460 genes scored (Top 5% of 9,207 expressed multi-isoform genes)

Author-reported evaluation · Source checked
Methods, coverage and source

Ladon, blood isoform direction accuracy, day0 14, top 5% of genes

ladon-20261115-protocol-blood-direction-day0-14

Aggregation: Not reported

Ladon: An RNA Foundation Model for Predicting Patient Outcomes (Zenodo PDF) · Page 27, paragraph beginning 'Finally, we analyzed 35 blood draws', value for 'Ladon' (over time)

Source checking is not independent reproduction. Release 2026-11-21-c8bea25e5abb.

Use this configuration

How it works, versions and access

Additional methodological details, versions and access terms have not yet been extracted into a structured profile for this configuration. Consult the linked sources and evaluation records below.

Evidence

Source checking verifies the cited claim or transcription. It does not establish independent reproduction.

Evidence table

Inspect claims, sources and review details

Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.

One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.

0 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-11-21-c8bea25e5abb
Property and statementOriginal source and locationReview and provenance

No evidence rows match these filters. Choose another scope or clear the search.

Sources and history

Release 2026-11-21-c8bea25e5abb · Record review: source checked

1 source record and release historyDownload this release (gzip)
Technical metadata and record identifiers

Technical metadata for this record is tracked under stable ID ladon-20261115-config-ladon-isoform-imputation.

Source locator: Page 28, Methods 'Model input'

Related records

How to cite

Ladon, pretrained encoder, isoform percentages imputed from gene-level counts. Rewire biological model benchmark database, release 2026-11-21-c8bea25e5abb. https://benchmarks.rewirebio.io/database/configuration/ladon-20261115-config-ladon-isoform-imputation/

Release 2026-11-21-c8bea25e5abb. Rankings and values change between releases, so keep the release id in the citation. Cite the original publication as well for any value you reuse: each result links its source.

The compilation and annotations made by Rewire in this release are offered under CC BY 4.0; values and text taken from cited sources stay under the terms of those sources and are not covered by this licence. Data licence

Suggest a correction