Back

LLM-PathwayCurator transforms enrichment terms into audit-gated decision-grade claims

Furudate, K.; Takahashi, K.

2026-02-19 bioinformatics
10.64898/2026.02.18.706381 bioRxiv
Show abstract

Pathway enrichment of omics data requires careful interpretation, and LLM-PathwayCurator transforms enrichment outputs into auditable, evidence-linked claims with redundancy-highlighting module mapping. Across seven cohorts from The Cancer Genome Atlas (TCGA), it achieved qualified coverage of 0.66 to 0.80 but fell to a range of 0.20 to 0.42 under context swap and 0.20 to 0.30 after supporting-gene dropout due to contract violations or weak gene support. LLM-PathwayCurator generates decision-grade claims with audit-gated abstention to establish a reproducible quality-assurance layer for omics interpretation.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

1
Nature Communications
○ 5641 papers in training set
Top 3%
30.6%
Similar papers in this journal
2
Nature Methods
○ 385 papers in training set
Top 1%
9.5%
Similar papers in this journal
3
Nature Genetics
○ 286 papers in training set
Top 0.7%
9.5%
Similar papers in this journal
4
Nature
○ 645 papers in training set
Top 2%
7.2%
Similar papers in this journal
50% of probability mass above
5
Genome Biology
○ 637 papers in training set
Top 3%
4.3%
Similar papers in this journal
6
Nucleic Acids Research
◐ 1281 papers in training set
Top 5%
4.0%
7
Nature Biotechnology
○ 172 papers in training set
Top 1%
3.4%
8
Cell Systems
○ 201 papers in training set
Top 2%
3.1%
9
Bioinformatics
◐ 1204 papers in training set
Top 6%
2.4%
10
Cell Genomics
○ 172 papers in training set
Top 2%
2.1%
11
eLife
● 5828 papers in training set
Top 52%
1.7%
12
Proceedings of the National Academy of Sciences
● 2444 papers in training set
Top 35%
1.1%
13
Nature Machine Intelligence
○ 70 papers in training set
Top 2%
1.1%
14
Genome Research
● 468 papers in training set
Top 5%
1.0%
15
Nature Aging
○ 60 papers in training set
Top 1%
1.0%
16
Genome Medicine
○ 183 papers in training set
Top 4%
1.0%
17
Nature Biomedical Engineering
○ 47 papers in training set
Top 2%
0.9%
18
Communications Biology
○ 993 papers in training set
Top 31%
0.8%
19
Nature Cell Biology
○ 118 papers in training set
Top 3%
0.8%
20
Molecular Systems Biology
○ 162 papers in training set
Top 3%
0.8%
21
Science Advances
● 1243 papers in training set
Top 31%
0.8%
22
Briefings in Bioinformatics
◐ 354 papers in training set
Top 8%
0.6%
23
Nature Microbiology
○ 155 papers in training set
Top 4%
0.6%
24
Nature Cancer
○ 39 papers in training set
Top 2%
0.6%
25
Cell Reports Methods
○ 165 papers in training set
Top 5%
0.6%

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.