ftINIT helpers #388

johan-gson · 2022-05-13T09:25:56Z

Main improvements in this PR:

*Added help functions that are used together with ftINIT in RAVEN. A pull request containing the ftINIT implementation has been done separately in RAVEN, but the changes should not be dependent on each other.

I hereby confirm that I have:

Tested my code on my own computer for running the model
Selected develop as a target branch

…man-GEM to RAVEN - these are used by ftINIT.

…ious tINIT since the RAVEN function has been tagged with '_legacy'.

…ed legacy. This makes the pull requests in RAVEN and Human-GEM independent. Let's fix this at some later point.

JonathanRob · 2022-05-20T15:08:38Z

code/tINIT/getAATripletReactions.m

+%   Usage: rxns=getAATripletReactions(model, onlyRxnsWithoutGPRs)
+%
+
+aminoAcidsCap = {'Alanine';'Arginine';'Asparagine';'Aspartate';'Cysteine';'Glutamine';'Glutamate';'Glycine';'Histidine';'Isoleucine';'Leucine';'Lysine';'Methionine';'Metheonine';'Phenylalanine';'Proline';'Serine';'Threonine';'Tryptophan';'Tyrosine';'Valine'};


~~Not sure what Metheonine is, but it can probably be removed (should only be 20 AAs)~~

Edit: I stand corrected, apparently it exists in the model. My bad.

So, the problem is that there are reactions where the amino acids are misspelled - that's why I have added those weird ones :)

How about submitting another PR that fixes the misspelling, and merging that one first?

Yeah sorry, I should have actually looked into it further before commenting.

Is there a reason to not code in all 22 amino acids?

So, the point of this function is to identify reactions that convert amino acid triplets back and forth, i.e. chains of three amino acids <=> three separate amino acids. It is a measure of optimization - these reactions are pretty pointless in most practical uses with tINIT. It is like 700 reactions or something like that. So I only added the amino acids that help identify those reactions.

JonathanRob · 2022-05-20T15:20:09Z

code/tINIT/getINITModel2.m

@@ -1,5 +1,7 @@
 function [model, metProduction, essentialRxnsForTasks, addedRxnsForTasks, deletedDeadEndRxns, deletedRxnsInINIT, taskReport] = getINITModel2(refModel, tissue, celltype, hpaData, arrayData, metabolomicsData, removeGenes, taskFile, useScoresForTasks, printReport, taskStructure, params, paramsFT)
 % getINITModel2
+%   This function is deprecated - we suggest that you instead use ftINIT in
+%   RAVEN Toolbox together with the helpers in this repo.


I have mixed feelings about this. This function is still (nearly) the original implementation of the algorithm, and is not necessarily "deprecated". Although ftINIT is faster, it uses a different approach (with additional assumptions and simplifications). So unless we can definitively show that ftINIT is a more accurate algorithm, I hesitate to "deprecate" the original tINIT.

The question is what to write here. Should we say that there is a newer version, but not call it deprecated? I have some reason to believe that ftINIT is performing better. I'm supplying a powerpoint that shows one test I did with this function and compared it to tINIT without allowing flux in both direction and without metabolite secretion. ftINIT in this test gave the same result as the one where these were not allowed, while the standard getINITModel2 produced something that is quite different, full of gaps. The task step of ftINIT typically fills in ~13 rxns, while the same step for getINITModel2 fills in ~55 rxns. So, I think that getINITModel2 has some problems, and that the problems with ftINIT are much smaller. But I have not tested it super much.
tINITTestInfo.pptx

Ok, very good points. I think it's fine to state that there is a newer version, but would avoid using the word "deprecated", since it sometimes implies that the function will eventually be removed (which is not the intent).

I agree that some of the behavior you've seen with ftINIT (other than shorter compute time) can be interpreted as a clear improvement in performance/accuracy.

Ok, I wrote that there is a new improved version of the function available in RAVEN, I have pushed a new commit regarding this.

JonathanRob · 2022-05-20T15:26:27Z

code/tINIT/prepHumanModelForftINIT.m

+    'MAR10063'; ...
+    'MAR10064'; ...
+    'MAR10065'; ...
+    'MAR13082'};


There is a lot of hard-coded items in this function, which I generally would like to avoid. But since I don't really have better suggestions, I think it's fine to leave like this.

mihai-sysbio · 2022-05-20T15:29:17Z

code/tINIT/prepHumanModelForftINIT.m

+    'MAR09818'};
+
+%reactions that just pool metabolites
+poolRxns = { ...


Pool reactions are reactions that have pool in the name, so these could be defined dynamically.

JonathanRob · 2022-05-20T15:30:02Z

code/tINIT/removeDrugReactions.m

+function outModel=removeDrugReactions(inModel)
+% outModel=removeDrugReactions(inModel)
+%
+% Removes reactions from Human-GEM related to drugs


I wonder if it would be easier to just remove all reactions that are in the Drug metabolism subsystem. I imagine it might remove a few additional reactions, but unlikely that they would be important for anything. Something to think about in the future.

Aha, I never thought about that.

As a review, I have following concerns to this function:

there is some commented lousy code

it doesn't appear to be generic enough for future use, thus might be okay to be hard coded inside prepHumanModelForftINIT

Yeah, I'm not sure that should be there - this is to show how I tested that the code works. Maybe we could clean that up.

…ious version of tINIT.

JonathanRob · 2022-05-24T05:15:28Z

Ok, I'm still going through some of the details on this PR, though so far it looks ok to me. But why are the yaml conversion tests failing?

johan-gson · 2022-05-24T06:21:01Z

Yeah, I don't get that either, I cannot see that I have done anything with that. What I understood from the error messages is that it fails to allocate some resource for running them?

mihai-sysbio · 2022-05-24T06:21:23Z

Ok, I'm still going through some of the details on this PR, though so far it looks ok to me. But why are the yaml conversion tests failing?

Now resolved, the Actions runner was offline.

…an-GEM into ftINITHelpers

johan-gson added 4 commits May 5, 2022 08:10

refactor: Moved removeLowScoreGenes.m and scoreComplexModel.m from Hu…

21d898e

…man-GEM to RAVEN - these are used by ftINIT.

style: commented getINITModel2 as deprecated

b0f334a

feat: Added Human-GEM related files for ftINIT. Also adapted the prev…

25eddbc

…ious tINIT since the RAVEN function has been tagged with '_legacy'.

fix: Changed back the getINITModel - it no longer calls something nam…

b4cc6ab

…ed legacy. This makes the pull requests in RAVEN and Human-GEM independent. Let's fix this at some later point.

haowang-bioinfo requested a review from JonathanRob May 15, 2022 01:43

JonathanRob reviewed May 20, 2022

View reviewed changes

mihai-sysbio changed the title ~~Ft init helpers~~ ftINIT helpers May 20, 2022

JonathanRob reviewed May 20, 2022

View reviewed changes

mihai-sysbio reviewed May 20, 2022

View reviewed changes

JonathanRob reviewed May 20, 2022

View reviewed changes

style: Removed the word "deprecated" from the description of the prev…

22c6008

…ious version of tINIT.

fix: skip adding metabolites step if none are to be added

f8b0269

JonathanRob approved these changes May 26, 2022

View reviewed changes

johan-gson added 2 commits June 6, 2022 11:09

style: Removed some commented out code

30c401f

Merge branch 'ftINITHelpers' of https://github.com/SysBioChalmers/Hum…

f47299c

…an-GEM into ftINITHelpers

haowang-bioinfo approved these changes Jun 13, 2022

View reviewed changes

haowang-bioinfo merged commit cf02955 into develop Jun 14, 2022

mihai-sysbio deleted the ftINITHelpers branch June 14, 2022 04:19

mihai-sysbio mentioned this pull request Jun 17, 2022

Human 1.12 #396

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ftINIT helpers #388

ftINIT helpers #388

johan-gson commented May 13, 2022

JonathanRob May 20, 2022 •

edited

Loading

johan-gson May 20, 2022

mihai-sysbio May 20, 2022

JonathanRob May 20, 2022

mihai-sysbio May 20, 2022 •

edited

Loading

johan-gson May 20, 2022

JonathanRob May 20, 2022

johan-gson May 20, 2022

JonathanRob May 21, 2022

johan-gson May 21, 2022

JonathanRob May 20, 2022

mihai-sysbio May 20, 2022

JonathanRob May 20, 2022

johan-gson May 20, 2022

haowang-bioinfo May 28, 2022

johan-gson May 28, 2022

haowang-bioinfo Jun 1, 2022

JonathanRob commented May 24, 2022

johan-gson commented May 24, 2022

mihai-sysbio commented May 24, 2022 •

edited

Loading

ftINIT helpers #388

ftINIT helpers #388

Conversation

johan-gson commented May 13, 2022

Main improvements in this PR:

JonathanRob May 20, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mihai-sysbio May 20, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

JonathanRob commented May 24, 2022

johan-gson commented May 24, 2022

mihai-sysbio commented May 24, 2022 • edited Loading

JonathanRob May 20, 2022 •

edited

Loading

mihai-sysbio May 20, 2022 •

edited

Loading

mihai-sysbio commented May 24, 2022 •

edited

Loading