microbialphenotypes / OMP-ontology

This repository contains the editors's working copy of OMP and public release files.
8 stars 2 forks source link

How to annotate Nichols' conflicting conditions? #236

Open peterwu19881230 opened 6 years ago

peterwu19881230 commented 6 years ago

In Nichols' 324 conditions, there are 114 unique chemicals (or treatments like UV). For each chemical there might be several concentrations. The following would cause issue ( because we will have increased resistance and decreased resistance for the same chemical):

Chemical Conc1: Reduced fitness score=0 Chemical Conc2: Reduced fitness score=-1 Chemical Conc3: Reduced fitness score=1

I have made a google sheet that has the first 20 treatments with various concentrations. If all reduced fitness scores are 0 in that particular treatment for that knockout, that row is removed. https://docs.google.com/spreadsheets/d/1V5StiD_QaXmxhnXfO3Cf-EML4OIDYRA8ujKCv68HCSY/edit#gid=754243435

(FYI: I can't make all the treatments in a file because there is a limitation for the R package: googlesheets -> I have tested that the maximum No. of tabs is 75.)

I have also made a similar file like the above one except it only extracts fitness scores where there are both 1 and -1 present in the conditions of that treatment. There are only 40, so I think we can resolve them case by case: https://docs.google.com/spreadsheets/d/19ZpYjQNXSquOGng69YrwbD19mfBEZaBUvsxytZOtnpw/edit

dsiegele commented 6 years ago

Hi Peter,

I'm confused by what you are planning to do. Can you explain it further?

I.) In the case where 3 increasing concentrations of a chemical/antibiotic are tested, what are the expected scores and what annotations will you make for each of them?

1) If a mutant has fitness scores 0,0,0, will there be an annotation? If yes, what will the annotation be? 2) If a mutant has fitness scores 1,0,0, what will the annotation be? 3) If a mutant has fitness scores 1,1,0, what will the annotation be? 4) If a mutant has fitness scores 1,1,1, what will the annotation be? 5) If a mutant has fitness scores 0,0, -1, what will the annotation be? 6) If a mutant has fitness scores 0,-1,-1, what will the annotation be? 7) If a mutant has fitness scores -1, -1, -1, what will the annotation be?

II.) I think the 3 conditions with SDS+EDTA need to be separated into 2 sets of comparisons:

1) SDS 0.5%/EDTA 0.1 mM vs SDS 0.5%/EDTA 0.5 mM

2) SDS 0.5%/EDTA 0.5 mM vs SDS 1.0%/EDTA 0.5 mm

III.) Above, you wrote "If all reduced fitness scores are 0 in that particular treatment for that knockout, that row is removed." I don't understand which row this refers to.

IV.) What does 'NA' mean? Does this mean there isn't a fitness scores for this combination of strain and condition in the table?

peterwu19881230 commented 6 years ago

I) A real example (in my second file the conflict doesn't exist for 3 concentration treatment. Let me use a four to explain):

Original name in Nichols' A22-0.5 A22-15.0 A22-2.0 A22-5.0 - ECK0165-GLND -1 1 1 0

ECK0165-GLND is going to be annotated by:

  1. decreased resistance under A22 (note: concentration: 0.5)

  2. increased resistance under A22 (note: concentration: 15.0 and 2.0) (p.s. 15 is out of order)

  3. We currently don't annotate "no phenotype"

  4. Annotate the 1

  5. Annotate the 1s (We can discuss whether to annotate once and have 2 concentrations in the note or 2 annotations with different concentrations in the note)

  6. Similar as 3.

  7. Annotate the -1

  8. Similar as 3

  9. Similar as 3

II) I don't quite understand this part. Aren't we comparing the scores to the imaginary wildtype? I remember we discussed in the lab meeting but I guess I don't fully understand what the "relative to" should be.

III) In both files I made, if scores across all concentrations for that treatment are all 0, I remove them => 0 0 0 <= We don't have to annotate this