502 error getting pubtator results for doc 10364520

[andrewsu]

https://mark2cure.org/document/pubtator/10364520.json reproducibly returns a 502 error

Other valid doc requests (eg https://mark2cure.org/document/pubtator/16344560.json) return just fine. Bad doc IDs (eg https://mark2cure.org/document/pubtator/1.json https://mark2cure.org/document/pubtator/ABC.json) correctly returning 404.

[gtsueng]

This might be more on pubtator's end? The browser version of pubtator doesn't show any annotations for this abstract.

[andrewsu]

Great point. Pubtator API indeed does not have any annotations for this document:

https://www.ncbi.nlm.nih.gov/CBBresearch/Lu/Demo/RESTful/tmTool.cgi/Chemical/10364520/JSON/
https://www.ncbi.nlm.nih.gov/CBBresearch/Lu/Demo/RESTful/tmTool.cgi/Gene/10364520/JSON/
https://www.ncbi.nlm.nih.gov/CBBresearch/Lu/Demo/RESTful/tmTool.cgi/BioConcept/10364520/JSON/

So I think there very well could be a pubtator error here, but M2C still should not return 502 when the set of annotations is empty...

[JTFouquier]

This issue is sporadic. This url Andrew posted (https://mark2cure.org/document/pubtator/10364520.json) mostly fails on dev, but once it did not. The only thing I did was add print statements to try to isolate the issue, so it wasn't a fix. Likely an issue with read_pubtator_bioc's as_writer OR the bioc_as_json, but was hard to isolate because it was producing random results for me.