Questions on job completion detection and litscan runtime

[Hengrui-Gu]

Thank you for your outstanding work on RNA summary generation! It has been very helpful for my research. I have two specific questions regarding the usage of your RNA summarizing API:

1. When I submit a job using the following command from the README file:
   curl -H "Content-Type:application/json" -d "{\"id\": \"bta-mir-16a\"}" localhost:8080/api/submit-job
   How can I determine when the summary for this RNA has been completed, so that I can safely submit the next job? Is there an endpoint or a recommended workflow for checking the job status?

2. How long does it generally take to perform a literature scan (litscan) for a single RNA? I tried submitting a large number of jobs in a relatively short period of time using the above command, but noticed that the summarizer keeps outputting “Polling for new jobs” in the logs. I am wondering if this means the generation step is stuck at the litscan literature search stage, or if there could be another bottleneck.

Thank you again for your great work. I look forward to your response!

[afg1]

Hi! Thanks for getting in touch, and I'm glad you find the tool useful!

1. The best way to check is the scan is completed is to look in the litscan_job table for an entry with the job ID you submitted. This is exactly what the polling does (see here). To see if a summary is completed, you will need to look in the litsumm_summaries table for a summary with the job_id you submitted. There isn't any polling in the application for this at the moment, but all job_ids should eventually get processed and end up with a summary there

2. This depends on the number of articles, since LitScan will parse the fulltext and extract sentences mentioning the RNA for each paper. This is ratelimited (by EuropePMC) to at most 10 articles/second, and will usually be quite a bit slower than that. For bta-mir-16a I see LitScan found 145 publications. This will take quite a while to process through LitScan, though the summary generation should be quick once it's done. I can't really give a concrete time per article unfortunately, but it can seem stuck for a while untill it suddenly isn't

Hopefully this helps, let me know if you need any further assistance!