Back

Quantification of circulating tumor DNA using deep learning

Jacobsen Skanderup, A.; Zhu, G.; Rahman, C.; Getty, V.; Baruah, P.; Carrie, H.; Lim, A.; Guo, Y. A.; Poh, Z.; Sim, N.; Abdelmoneim, A.; Cai, Y.; Ho, D.; Thangaraju, S.; Poon, P.; Lau, Y.; Gan, A.; Ng, S.; Odinokov, D.; Koo, S.-L.; Chong, D.; Tay, B.; Tan, T.; Yap, Y.; Chok, A.; Ng, M.; Tan, P.; Tan, D.; Wong, L.; Wong, P.; Tan, I.

2023-07-28 bioinformatics
10.1101/2023.07.28.550922 bioRxiv
Show abstract

Quantification of circulating tumor DNA (ctDNA) levels in blood enables non-invasive surveillance of cancer progression. Fragle is an ultra-fast deep learning-based method for ctDNA quantification directly from cell-free DNA fragment length profiles. We developed Fragle using low-pass whole genome sequence (lpWGS) data from multiple cancer types and healthy control cohorts, demonstrating high accuracy, and improved lower limit of detection in independent cohorts as compared to existing tumor-naive methods. Uniquely, Fragle is also compatible with targeted sequencing data, exhibiting high accuracy across both research and commercial targeted gene panels. We used this method to study longitudinal plasma samples from colorectal cancer patients, identifying strong concordance of ctDNA dynamics and treatment response. Furthermore, prediction of minimal residual disease in resected lung cancer patients demonstrated significant risk stratification beyond a tumor-naive gene panel. Overall, Fragle is a versatile, fast, and accurate method for ctDNA quantification with potential for broad clinical utility.

Matching journals

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

1
Nature Communications
○ 5641 papers in training set
Top 4%
26.7%
Similar papers in this journal
2
Genome Medicine
○ 183 papers in training set
Top 0.1%
13.2%
Similar papers in this journal
3
Genome Biology
○ 637 papers in training set
Top 1%
9.7%
Similar papers in this journal
4
Nature Biotechnology
○ 172 papers in training set
Top 0.6%
5.5%
Similar papers in this journal
50% of probability mass above
5
Bioinformatics
◐ 1204 papers in training set
Top 4%
4.4%
Similar papers in this journal
6
Nucleic Acids Research
◐ 1281 papers in training set
Top 5%
3.2%
7
Nature Methods
○ 385 papers in training set
Top 3%
2.8%
8
Communications Biology
○ 993 papers in training set
Top 12%
1.9%
9
Science Advances
● 1243 papers in training set
Top 18%
1.9%
10
eLife
● 5828 papers in training set
Top 50%
1.7%
11
Cell Reports Methods
○ 165 papers in training set
Top 2%
1.5%
12
Briefings in Bioinformatics
◐ 354 papers in training set
Top 5%
1.4%
13
Genome Research
● 468 papers in training set
Top 4%
1.3%
14
Nature Biomedical Engineering
○ 47 papers in training set
Top 1%
1.3%
15
iScience
○ 1154 papers in training set
Top 26%
1.1%
16
Cancer Research Communications
● 51 papers in training set
Top 1%
1.1%
17
Nature Genetics
○ 286 papers in training set
Top 4%
1.1%
18
PLOS Computational Biology
● 1863 papers in training set
Top 17%
1.1%
19
BMC Bioinformatics
○ 457 papers in training set
Top 5%
1.1%
20
Cell Reports Medicine
○ 153 papers in training set
Top 4%
1.0%
21
Advanced Science
○ 286 papers in training set
Top 8%
0.9%
22
Proceedings of the National Academy of Sciences
● 2444 papers in training set
Top 41%
0.8%
23
GigaScience
◐ 212 papers in training set
Top 4%
0.8%
24
Scientific Reports
○ 3612 papers in training set
Top 72%
0.8%
25
Cell Genomics
○ 172 papers in training set
Top 5%
0.6%
26
Clinical Cancer Research
● 64 papers in training set
Top 2%
0.6%
27
Cancer Research
● 130 papers in training set
Top 3%
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.