search

waldronlab / BugSigDB

A microbial signatures database
https://bugsigdb.org
7 stars 6 forks source link

testing by curators #11

Closed lwaldron closed 4 years ago

lwaldron commented 4 years ago

I've drafted the following Slack message to send to our curators. @tosfos shall I send it now or do you want me to wait, and is there anything you want me to change or add?

@channel I’ve just made accounts for all of you at https://bugsigdb.org. If you have a gmail account (which I used when I had them), it is probably in your spam folder. Those of you curating signatures, it would be great for you to try entering or editing an existing signature and make note of any difficulties you have in response to this message. It is not in a finished appearance yet, we are still focusing on data entry, storage, and performance. Some notes:

  1. the entire site is a “sandbox” now so you can enter, edit, and delete anything you want, and it will all be over-written anyways once we finish testing mode and move the current data over. Do your worst to try to break it, and don’t worry about messing up the data that are there.
  2. The process for a new study starts with “Add a Study” at the top of the page, where you’ll only have to enter study design and PMID. Then “Add Experiment” where you enter most details like body site, sequencing, statistical method, cutoffs, group names, and alpha diversity. Then finally, “Add Signature” where you only enter taxonomic identifiers and whether it is greater or lower abundance in Group 1. You should be able to find studies/experiments/signatures already in the database (most of the ones in the spreadsheet) by searching for PMID or keywords. Here’s an example of a previously completed study (The tongue microbiome in healthy subjects and patients with intra-oral halitosis): https://bugsigdb.org/Study_14
  3. When entering taxa, don’t do the “g__” etc, just start typing the name of the taxa whatever level of the taxonomy it is. If it doesn’t autocomplete, you need to look it up at https://www.ncbi.nlm.nih.gov/taxonomy then enter the numeric taxID you find there.
  4. Adding and editing facilities may not behave in the same way yet, make a note of what you are doing when you see an issue. We want to hear also anything you notice that would decrease your efficiency.
  5. You can either file issues at https://github.com/waldronlab/BugSigDB/issues or just reply here, up to you.
tosfos commented 4 years ago

It looks OK to me. I would just add a warning that the site is in active development and may misbehave in unpredictable ways.

lwaldron commented 4 years ago

Comments from Shaimaa Elsafoury:

  1. I didn’t find Cesarean section in the condition section.
  2. I also realized that statistical test section is a free form section which is not the case in our spreadsheet
  3. in alpha diversity section, what should we choose if the study didn’t conduct one of the tests.
  4. I also didn’t finish the data entry, but I saved my work in case something happened. However, I am not able to return back again to complete the data entry.
lwaldron commented 4 years ago

Comments from Fatima Zohra:

  1. Wasn't sure what to do when LDA Score is irrelevant (tried putting N/A and this was invalid) - add a note to tooltip?
  2. Wasn't sure what to do when there is only one variable region (add to tooltip, make upper and lower the same)
  3. Wasn't sure the difference between blank and unknown under Alpha Diversity - tooltip
  4. Thought "summary" belonged to Alpha Diversity, didn't realize it meant a summary of the edit. Multiple curators thought this meant some summary of the study or experiment.
lwaldron commented 4 years ago

Other feedback from discussion:

  1. tedious having to re-enter all data for nearly identical experiments, e.g. host species is always the same. Would be convenient only having to enter common values only once, either on their own forms or by copying previous experiment
  2. Were unclear that Alpha diversity "increased" / "decreased" meant in cases relative to controls
  3. For an experiment page, click "Subjects" [Edit] - it sends you to Editing Template:Experiments (section) which is template code, but user wanted the form again
  4. Not clear how to view and track the history of a study and its associated experiments and signatures
lwaldron commented 4 years ago

These comments are just copied in as made, and there's a lot to unpack. I'll work on organizing this feedback and commenting.

lgeistlinger commented 4 years ago

Some points from my side:

Summary / Minor edit / Watch page: what is the purpose of these fields on the bottom of the edit form?

We can remove these if it is confusing. See: https://meta.wikimedia.org/wiki/Help:Edit_summary https://meta.wikimedia.org/wiki/Help:Minor_edit https://www.mediawiki.org/wiki/Help:Watching_pages

And I wonder whether we should indeed remove these fields or do we consider them beneficial?

tosfos commented 4 years ago

maybe prefill when adding an experiment to a study with the values of an already existing experiment of the study? Maybe we could even allow to select the experiment for pre-filling the fields when adding a new experiment to a study that already has > 1 experiments

We can add a button on each experiment that says "Duplicate this Experiment" or something like that and it would bring up a preloaded form with those values.

lgeistlinger commented 4 years ago

I like that. As an alternative, I could imagine that adding an experiment becomes a two-step procedure: 1. form to select whether you want pre-filling from an existing experiment of the study, 2. current form where to enter / change the info of an experiment, pre-filled with the information of experiment selected in step 1.

lwaldron commented 4 years ago

Pasting in feedback from Lucille Mellor. @lgeistlinger, would you copy and organize all this feedback into the wiki, where we can sort and prioritize, then we should return to using the issue tracker for discrete issues.

1) If a study sequenced 2 difference variable regions instead of a range, there does not appear to be a way to enter this (i.e. V1&V6 instead of V1-V6) 2) For the statistical analysis field, it would be helpful if there was a drop-down list available to view available options (there appears to be a controlled dictionary for this field but also free-text can be entered) 3) Is there a way to delete a study or experiment entry? For example, if you enter the wrong PMID in error, there does not appear to be a way to delete the study from the database (if you edit the erroneous PMID, the new study adds but the original study remains as well) 4) For the example you provided on the tongue microbiome, there is a summary of the experiments entered on the main study page, but I don't see a summary like this for my test studies unless I look at the Recent Changes tab. This summary is helpful for seeing what you've entered so far 5) I don't see an option to add signatures for the experiments. Is this functionality available yet?

lgeistlinger commented 4 years ago

Alright, will do.

lwaldron commented 4 years ago

Quick note about Lucille's point 1 - I think that's a very unusual technique (that paper used a microarray, not sequencing), and it's probably not worth making an effort to accommodate such an unusual case. My advice for that paper will be just to put V1-V6 as a hack, and the fact the "sequencing method" is microarray will set it apart methods-wise anyways.

lwaldron commented 4 years ago

I'm closing this issue now, instead working on organizing feedback at https://github.com/waldronlab/BugSigDB/wiki/Feedback-from-curators.