ShimaaBetah
Protein-Protein Interaction Network Usecase
Greetings, The potential of GraphoPlex being an effective tool in Pharmaceutical industries is immense. However, I have some concerns, along with feature requests.
Firstly, you can integrate GraphoPlex with StringDB via their API. You can import TSV files for GeneOntology Biological Processes into the database. For example, the biological process of the regulation of ATP-dependent activity that has +23k annotations, the evidence score for each protein interaction is calculated by StringDB, based on co-expression, Association in Curated Databases, Experimental/Biochemical Data, or co-mentioning in Pubmed abstracts. This integration with easy import/export methods can help bioinformatics and pharmaceutical researchers to analyze Protein-Protein Interaction Networks for drug development.
Secondly, I suggest adding extensions/plugins for k-means/DBSCAN/MCL clustering methods, a non-negligible percentage of GraphoPlex usecases would need clustering algorithms to be embedded to the server.
Thirdly, I have some concerns about the round-trip latency for the database in general. As shown in Thesis Page27, the total execution time is a bit concerning. I would suggest minimizing the usage of HTTP(S) protocol, along with TCP Protocol optimizations (Merely Beneficial example). I can contribute to GraphoPlex to comply with ISO25010 standard, especially the functional correctness part.
As a final note, I have made relevant projects in Java, I think you might be interested in them:
- GCKit, A research project to optimally cluster Protein-Protein Interaction networks, along with Weighted Clustering Coefficient for a sub-graph, and multi-threaded calculation for a custom power set for the graph.
- SparkDB, A database project to store CSV tables in memory, and manipulate them easily.
Best regards.
