TY  - GEN
N2  - Genomic screening is an increasingly important part of cancer care. Screening often detects somatic variants in the tumor sample of varied clinical significance: some are well understood, some are connected to only loose evidence, and some are unknown altogether. The purpose of this project was to construct a tool that could present state-of-the-art pathway information to genomics experts evaluating the clinical significance of variants. A web service was created that runs queries against the Reactome pathway database in search of common pathway activity between variants of a clinical case that are known to be pathogenic and those that are of unknown significance. It was integrated into the software infrastructure of a high-volume genomics lab at Providence St. Joseph?s Health. A very different approach to the same problem was attempted via the cloud database product Google BigQuery. The project so far has failed to be of clinical utility.  Two areas of improvement could remedy that situation in future iterations: a more stable network visualization technique, and higher resolution mapping of novel variants to pathway databases via accounting for the effect of alterations on particular protein subdomains.  In its current form the project was not able to reap the benefits of a graph database in particular.  A simpler focus on a small number of ?canonical pathways? looks like a quicker path to a value-added user interface.
DO  - 10.6083/fq977v472
DO  - DOI
AB  - Genomic screening is an increasingly important part of cancer care. Screening often detects somatic variants in the tumor sample of varied clinical significance: some are well understood, some are connected to only loose evidence, and some are unknown altogether. The purpose of this project was to construct a tool that could present state-of-the-art pathway information to genomics experts evaluating the clinical significance of variants. A web service was created that runs queries against the Reactome pathway database in search of common pathway activity between variants of a clinical case that are known to be pathogenic and those that are of unknown significance. It was integrated into the software infrastructure of a high-volume genomics lab at Providence St. Joseph?s Health. A very different approach to the same problem was attempted via the cloud database product Google BigQuery. The project so far has failed to be of clinical utility.  Two areas of improvement could remedy that situation in future iterations: a more stable network visualization technique, and higher resolution mapping of novel variants to pathway databases via accounting for the effect of alterations on particular protein subdomains.  In its current form the project was not able to reap the benefits of a graph database in particular.  A simpler focus on a small number of ?canonical pathways? looks like a quicker path to a value-added user interface.
AD  - Oregon Health and Science University
T1  - VarGraph: a decision support tool for variant classification using pathway databases
DA  - 2019
AU  - Ball, David
L1  - https://digitalcollections.ohsu.edu/record/7585/files/ball.david._2019.pdf
PB  - Oregon Health and Science University
PB  - Oregon Health and Science University
PY  - 2019
ID  - 7585
L4  - https://digitalcollections.ohsu.edu/record/7585/files/ball.david._2019.pdf
KW  - Genomics
KW  - Software
KW  - Neoplasms
KW  - databases
KW  - genetic
TI  - VarGraph: a decision support tool for variant classification using pathway databases
Y1  - 2019
L2  - https://digitalcollections.ohsu.edu/record/7585/files/ball.david._2019.pdf
LK  - https://digitalcollections.ohsu.edu/record/7585/files/ball.david._2019.pdf
UR  - https://digitalcollections.ohsu.edu/record/7585/files/ball.david._2019.pdf
ER  -