Question
I have a counts matrix of single cell transcriptomic data, cell.proportions.q1.mtx.gz. The data contains measurements from 5350 cells, each with 1000 measured genes. That data was generated by measuring data from a co-culture of a few unique cell types. Can you tell me the proportions of those cells in descending order to the nearest 5 percent? Drop any cells with questionable QC. Expected output: 35;35;20;10 if there are 4 cell types.
- Predicted answer
50;35;15Most likely answer for this question; not uniquely determined from submitted scores.
- Domain
- Single-cell
- Question style
- Synthetic/Augmented Data
An answer can be uniquely inferred for 54 of 100 questions from submitted scores. Green: correct · Red: incorrect. For questions without a uniquely determined answer: Blue: matches prediction · Orange: differs from prediction. These are inferred answer checks, not official per-question grades.
Galaxy
12/12 terminal item results
UDT = user-defined tool: custom code executed by Galaxy. Route badges summarize analysis tools, excluding uploads and automatic conversions. How usage is counted.
Codex GPT-5.5
3 replicates
Codex GPT-5.5
Replicate 1
Matches predictionUDT-only
Answer50;35;15
WorkflowSC Cell Type Proportions→SC Cell Type Proportions v2→SC Cell Type Proportions v3→SC Cell Type Proportions v4
- Total tokens
- 577.6K
- Output
- 11.3K
Codex GPT-5.5
Replicate 2
Matches predictionUDT + standard tools
Answer50;35;15
WorkflowSingle-cell proportions diagnostic→Single-cell proportions diagnostic v2→UDT command field probe→UDT shell probe→UDT shell probe v2→1 more tools
- Total tokens
- 6.4M
- Output
- 33.2K
Codex GPT-5.5
Replicate 3
Matches predictionUDT + standard tools
Answer50;35;15
WorkflowSingle-cell proportions diagnostic→Single-cell proportions diagnostic v2→UDT JS probe→UDT JS probe v2→Anndata Import→6 more tools
- Total tokens
- 5.7M
- Output
- 22.2K
Codex GPT-5.6 Sol
3 replicates
Codex GPT-5.6 Sol
Replicate 1
Matches predictionStandard tools only
Answer50;35;15
WorkflowAnndata Import→Anndata Inspect→Anndata Manipulate→Scanpy Inspect→Datamash Ops→5 more tools
Codex GPT-5.6 Sol
Replicate 2
Matches predictionUDT-only
Answer50;35;15
WorkflowSingle-cell matrix QC diagnostic→Single-cell QC and cluster exploration→Final QC-filtered single-cell proportions
- Total tokens
- 670.5K
- Output
- 8.8K
Codex GPT-5.6 Sol
Replicate 3
Matches predictionUDT-only
Answer50;35;15
WorkflowSingle-cell proportions diagnostic→Validated single-cell type proportions
- Total tokens
- 2.6M
- Output
- 11.8K
Codex + DeepSeek-v4-pro-0813
3 replicates
Codex + DeepSeek-v4-pro-0813
Replicate 1
RerunMatches predictionUDT + standard tools
Answer50;35;15
WorkflowUser-defined tool → Summary Statistics1 → anndata import → scanpy filter cells
- Total tokens
- 5.1M
- Output
- 43.5K
Codex + DeepSeek-v4-pro-0813
Replicate 2
Matches predictionUDT + standard tools
Answer50;35;15
WorkflowNo Galaxy tool job recorded
- Total tokens
- 64.2M
- Output
- 167.0K
Codex + DeepSeek-v4-pro-0813
Replicate 3
RerunMatches predictionUDT + standard tools
Answer50;35;15
Workflowanndata import → anndata inspect → anndata manipulate → scanpy inspect → seurat create → anndata export → User-defined tool
- Total tokens
- 8.4M
- Output
- 47.9K
Codex GPT-5.6 Luna
3 replicates
Codex GPT-5.6 Luna
Replicate 1
Matches predictionUDT-only
Answer50;35;15
WorkflowSingle-cell Proportion Diagnostic→Single-cell Proportion Validation→Single-cell Proportions Final
- Total tokens
- 1.1M
- Output
- 23.0K
Codex GPT-5.6 Luna
Replicate 2
Matches predictionUDT-only
Answer50;35;15
WorkflowMatrix runtime probe→Explore single-cell mixture→Quick matrix stats→Probe single-cell mixture clusters→Validate single-cell mixture clusters→3 more tools
- Total tokens
- 1.5M
- Output
- 33.1K
Codex GPT-5.6 Luna
Replicate 3
Matches predictionStandard tools only
Answer50;35;15
WorkflowAnndata Import→Anndata Inspect→Anndata Manipulate→Scanpy Inspect→Datamash Ops→5 more tools
- Total tokens
- 20.5M
- Output
- 35.8K