Closed ftzohra22 closed 1 year ago
@tosfos just noticed this is an issue for all the studies with multiple experiments
It looks like this field is actually working fine. It's some of the other ones that are problematic. We did implement difference checking for the Experiments that works as follows:
If a certain field in Experiment 2 has the same value as in Experiment 1, it should not be displayed. We will always set the semantic property, so that it can be queried and show up in the CSV exports.
The exception is for all fields in the Alpha Diversity section, which should always be displayed on all Experiments, as requested in #120.
We'll work on fixing this. Please let me know if any of the above is incorrect, @lwaldron @lgeistlinger .
Please review:
All unchanged values in Experiment 2 are being hidden. We are showing all changes in Alpha Diversity no matter what, but in this Study's Experiment 2, only Chao1 is filled in.
If this looks OK, we'll apply it to production.
For comparison, here is the same page in production: https://bugsigdb.org/Study_696
Sorry I missed this earlier. Your proposal looks good @tosfos , showing only the differences (except for alpha diversity) for subsequent experiments makes it much easier to see how each experiment differs.
This has been fixed. You will see decreased site performance for a bit.
We did notice one issue that should be discussed. Sometimes certain Experiments have fields with no value, and this results in an output that might not be expected.
Please check Study 750, MHT correction. In this Experiment, we get an interesting output because Experiment 3 does not have this field set.
Although the Experiment 3 MHT correction printout follows the logic (it is shown because by being blank it is not equal to the preceding value), the printout without any value looks confusing. Would it be helpful to output Not specified or something similar? Removal of empty parameters looks worse as it can be interpreted as "unchanged". Or are unset values an error of omission that should default to a certain value?
We could:
Option 3 seems the easiest and just fine, not requiring replacing currently blank values with "not specified". Even the status quo doesn't seem bad if there's much work involved, since this should be a relatively rare curation error. That particular example in Study 750 it is a curation error, and I doubt a study would ever specify that they used MHT for the first two experiments, then changed strategy and didn't specify for the third (and it would also be odd to change from yes to no between experiments in the same paper).
We implemented option 3.
Can this be closed?
Yes, looks good.
https://bugsigdb.org/Study_750 has more than one experiment and the variable region is not showing under Experiment 1. The information is entered but does not appear to be saving and showing on the saved pages for the subsequent experiments.