Back

SKiM - A generalized literature-based discovery system for uncovering novel biomedical knowledge from PubMed

Raja, K.; Steill, J.; Ross, I.; Tsoi, L. C.; Kuusisto, F.; Ni, Z.; Livny, M.; Thomson, J.; Stewart, R.

2020-10-17 bioinformatics
10.1101/2020.10.16.343012 bioRxiv
Show abstract

Literature-based discovery (LBD) uncovers undiscovered public knowledge by linking terms A to C via a B intermediate. Existing LBD systems are limited to process certain A, B, and C terms, and many are not maintained. We present SKiM (Serial KinderMiner), a generalized LBD system for processing any combination of A, Bs, and Cs. We evaluate SKiM via the rediscovery of discoveries by Don Swanson, who pioneered LBD. Using only literature from the 19th century up to a year before Swansons discoveries, SKiM uncovers all five discoveries. We apply SKiM to repurposing drugs for 26 conditions of high prevalence. Manual analysis confirmed 65 discoveries useful for four diseases from Swansons discoveries from one to 31 years prior to their first validation by clinical trials. SKiM predicts many new potential drug candidates representing prime targets for wet lab validation. SKiM can be applied to any biomedical inquiry sufficiently mentioned in the literature.

Matching journals

The top 10 journals account for 50% of the predicted probability mass.

1
Scientific Reports
○ 3612 papers in training set
Top 7%
7.8%
Similar papers in this journal
2
Nature Communications
○ 5641 papers in training set
Top 21%
7.8%
Similar papers in this journal
3
Bioinformatics
◐ 1204 papers in training set
Top 3%
7.8%
Similar papers in this journal
4
Scientific Data
○ 209 papers in training set
Top 0.5%
4.8%
Similar papers in this journal
5
Briefings in Bioinformatics
◐ 354 papers in training set
Top 2%
4.8%
Similar papers in this journal
6
iScience
○ 1154 papers in training set
Top 4%
4.3%
7
npj Digital Medicine
○ 118 papers in training set
Top 1%
4.0%
8
Cell Genomics
○ 172 papers in training set
Top 1%
3.4%
9
Database
◐ 61 papers in training set
Top 0.2%
2.8%
10
Genomics, Proteomics & Bioinformatics
188 papers in training set
Top 0.7%
2.8%
50% of probability mass above
11
Patterns
○ 78 papers in training set
Top 0.8%
2.6%
12
Bioinformatics Advances
◐ 203 papers in training set
Top 2%
2.6%
13
PLOS Computational Biology
● 1863 papers in training set
Top 12%
2.4%
14
Nucleic Acids Research
◐ 1281 papers in training set
Top 7%
2.4%
15
Communications Medicine
○ 113 papers in training set
Top 2%
1.7%
16
eLife
● 5828 papers in training set
Top 54%
1.5%
17
eBioMedicine
○ 183 papers in training set
Top 3%
1.5%
18
Proceedings of the National Academy of Sciences
● 2444 papers in training set
Top 31%
1.4%
19
npj Systems Biology and Applications
○ 125 papers in training set
Top 1%
1.4%
20
Computational and Structural Biotechnology Journal
● 242 papers in training set
Top 4%
1.3%
21
PLOS ONE
● 5266 papers in training set
Top 53%
1.1%
22
BMC Bioinformatics
○ 457 papers in training set
Top 5%
1.1%
23
Nature Methods
○ 385 papers in training set
Top 5%
1.1%
24
The Pharmacogenomics Journal
○ 11 papers in training set
Top 0.1%
1.1%
25
Clinical Pharmacology & Therapeutics
○ 25 papers in training set
Top 0.3%
1.1%
26
NAR Genomics and Bioinformatics
◐ 242 papers in training set
Top 3%
1.1%
27
Genome Biology
○ 637 papers in training set
Top 8%
1.1%
28
Cell Systems
○ 201 papers in training set
Top 4%
1.0%
29
Nature
○ 645 papers in training set
Top 9%
1.0%
30
Journal of Cheminformatics
○ 29 papers in training set
Top 0.6%
1.0%

"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.