Egem2 by fanela · Pull Request #50 · sriram-lab/egem

fanela · 2019-09-25T21:19:12Z

No description provided.

This reverts commit da8f33c.

…anges before starting new branch.

…o make scatter plots and calculate correlation in MATLAB_CODE_me. Saved outputs constrain_flux_regulation in VariablesSaved. Renamed min variable to min_model to avoid conflict with min function.

… values. Saved resulting figures.

…into structures called sigRxn... while long program ran on desktop.

…n PC, one on desktop.

…alue. compared 1E-1 FBA and 1E-1 fBA and saved variables.�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D�[D

…uxes_allrxns for all 6 weights.

…into local repo.

… positive or negative.

ScottCampit · 2019-09-26T01:52:27Z

matlab/MeCorr/MATLAB_CODE_allrxns.m

+            offgenes = unique(ccleids_met(ccle_expression_metz(:,iii) < -2));
+
+            %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+            %  set the glucose uptake based on media


These lower bounds only change glucose abundance for 4 medium conditions. The code matlab/scripts/nutrient_sensitivty/medium_LB_constraints.m in master can improve your drug sensitivity predictions.

ScottCampit · 2019-09-26T01:55:33Z

matlab/MeCorr/MATLAB_CODE_allrxns.m

+load supplementary_software_code hdacexpfcs hdacexpallgeneids 
+%contains gene expression data after treatment with hdac inhibitors
+
+% Converting arrays to tables and renaming some Variables to make


Once you prettify these variables, you should save them in vars and just load them rather than formatting them during every iteration within this script.

ScottCampit

@fanela, there are several changes that need to be made before pushing this to the master branch. General stuff that I saw:

Variable names are still super confusing - please help make the code readable by changing the variables names to something more understandable (ie instead of using v1, use KMTi_auc directly, which is much more clear.
There's a ton of code that is repetitive between the three new scripts in the MeCorr directory. Since you're reusing these variables, save them as .mat files and load them at the beginning of each script. That way, it's easier to figure out what this script is actually doing.
I found a few bugs that should be addressed and that should change your results and potentially your interpretation of the results.

Also, I moved some of your old matlab scripts in the drug_sensitivity folder. I though MeCorr was too generic, because there are several modules that compute correlation values between expression and Me flux. If you can, move these scripts there and save the new matlab variables in the vars directory.

ScottCampit · 2019-09-26T01:57:13Z

matlab/MeCorr/MATLAB_CODE_allrxns.m

+    fluxes_allrxns= NaN(height(exptidcelllinemediamatch),length(model2.rxns));
+    g_rate= NaN(height(exptidcelllinemediamatch), length(model2.rxns));
+
+    for i = 1:height(exptidcelllinemediamatch)


Can you change i, ii, and iii into better variable names that are more accurate? It becomes really confusing reading the rest of the code

Repeat this for a lot of the variables below. Our goal is to have this code more readable so that it is easy to figure out what's going on without comments.

ScottCampit · 2019-09-26T02:06:25Z

matlab/MeCorr/MATLAB_CODE_allrxns.m

@@ -0,0 +1,419 @@
+      %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


Rename the three files with MATLAB_CODE_* files - it's not clear from the get-go what the difference between allrxns, me, and methyl are without opening and carefully reading the file contents.

ScottCampit · 2019-09-26T02:12:37Z

matlab/MeCorr/MATLAB_CODE_allrxns.m

+        ii = find(ismember(ctd2celllineidname_id, exptidcelllinemediamatch(i,2)));
+        iii = find(ismember(celllinenames_ccle1, ctd2celllineidname(ii,1)));
+        if ~isempty(iii)
+            iii  = iii(1);


Why is iii set to the first value if it's not empty? If length(iii) > 1, then you may want to average ccle_expression_metz for those columns, then get ongenes and offgenes. This gets the average gene expression values for the duplicated cell line.

ScottCampit · 2019-09-26T02:15:20Z

matlab/MeCorr/MATLAB_CODE_allrxns.m

+rho_p= NaN(height(hmei_list),length(model2.rxns));
+for j = 1:height(hmei_list)
+    fx = find(ismember(ctd2compoundidname_name, hmei_list(j,1)));
+    ix = ismember(drug_auc_expt(:,3), ctd2compoundidname_id(fx,1)); sum(ix);


separate code after semicolon to new line to improve readability

ScottCampit · 2019-09-26T02:16:50Z

matlab/MeCorr/MATLAB_CODE_allrxns.m

+%% Create Table of Significant Reactions by Correlation value
+% Workflow: Change threshold. Change struct field name (e.g. above3)
+% accordingly
+sigTF= (abs(rho) > 0.3);


Why are you taking the absolute value of rho? Negative values are important and have a statistical interpretation.

Also if you want to get significant reactions, you would want to filter using the p-values as well as R value. That is, you want the reaction to have an R-value >= 0.3 and a p-value < 0.05.

Line 195: I coded abs(rho) > 0.3 so that the negative correlations below -0.3 would be selected as well. I am including those correlations in downstream analysis.

I see, you should create two variables have rho > 0.3 and rho < -0.3, unless you wrote in a way to distinguish between the negatively and positively correlated entries by saving their respective indices first. That way, you can combine them later to make a plot while keeping the respective scalar directions.

ScottCampit · 2019-09-26T02:31:44Z

matlab/MeCorr/MATLAB_CODE_methyl.m


-
 [ix, aapos] = ismember( mediareactions1(20:37),acetylation_model.rxns);
 posgluc = 1385;  % glucose uptake reaction in recon1.


Don't hard-code this - use find(ismember(model.rxns, 'EX__glc_D_e')) to find the glucose exchange reaction. Same for glutamine and linoleic acid.

ScottCampit · 2019-09-26T02:32:35Z

matlab/MeCorr/MATLAB_CODE_methyl.m


-
-    model3.c(rxnpos) = epsilon_acetylation;
+    model3.c(rxnpos) = epsilon_methylation;


ScottCampit · 2019-09-26T02:34:00Z

matlab/MeCorr/MATLAB_CODE_methyl.m

-
-
-    model3.c(rxnpos) = epsilon_acetylation;
+    model3.c(rxnpos) = epsilon_methylation;


Also repetitive - model3.c should not change the way your code is written.

ScottCampit · 2019-09-26T02:35:15Z

matlab/MeCorr/MATLAB_CODE_methyl.m

- acetgenessorted = unqgenes(spos);
+    
+    [sx spos] = sort(acet_screen_rpmi(:,1));
+    acetgenessorted = unqgenes(spos);


change variable names to something more generic since this code looks at methylation now

ScottCampit · 2019-09-26T02:37:38Z

matlab/MeCorr/MATLAB_CODE_methyl.m


        [solf.x,sol11] =  constrain_flux_regulation(model3,[],[],0,0,0,[],[],minfluxflag);
        if ~isempty(solf.x) && ~isnan(solf.x)
            acet_screen_rpmi(i,2) = solf.x(objpos); % impact on growth


Change variable name so it's more obvious that it's the growth rate from the constrained metabolic model. Also, are you intentionally using the acetylation model? It is confusing since you're storing the objective coefficient value into the variable name epsilon_methylation.

fanela added 30 commits June 28, 2019 13:50

Merge branch 'master' into egem-lf

0253a3e

Removed 2 figures that didn't belong in scripts folder

addc226

Round a workaround for deleteModelGenes error due to min model

e562993

Made bulk methylation model compatible. Added 2 reactions to min.

1b0a576

Pulled master. Modified MATLAB_CODE_methyl & methylVariables

28ff648

Forgot to delete all conflict markers. Now they're deleted.

cbe6502

Shortened names of variables that were long (4 variables)

4d3a40f

Save work before revert

da8f33c

Revert "Save work before revert"

1cc18d8

This reverts commit da8f33c.

Discovered MATLAB_CODE used data from Ludlow 2016 not 2015. Saving ch…

ca73666

…anges before starting new branch.

Using Rees 2016 data. Changed list of inhibitors to Me inhibitors

9fe4d5e

Moved block of code to beginning of last module. Aligned variable names.

0efd35a

MATLAB_CODE_me contains only last module of MATLAB_CODE. Wrote code t…

9722468

…o make scatter plots and calculate correlation in MATLAB_CODE_me. Saved outputs constrain_flux_regulation in VariablesSaved. Renamed min variable to min_model to avoid conflict with min function.

Put all folders of figures into 1 folder

b6bded9

Ran min_model through MATLAB_CODE_me using 3 more epsilon_methylation…

7a21ecd

… values. Saved resulting figures.

Attempt to push on Fri 7/12

b4e0db1

First time pushing to different remote: egem-lf

ed8d0a8

2nd or 3rd commit after hiatus from GitHub. Today, formatted results …

5f11c16

…into structures called sigRxn... while long program ran on desktop.

Saved significant reactions of fluxes_allrxns output

34c1b31

From Sriram lab's desktop. Saved new variable grate_1FBA_allRxns.

1d4617f

Merged 2 local branches of egem2�[D�[D�[D�[D�[Dremote �[Begem2; one o…

aaef790

…n PC, one on desktop.

Saved significant reactions of 4 mil loop

422e738

Fixed some variables. Ran a few more simulations, and now have FBA fl…

f5336a6

…uxes_allrxns for all 6 weights.

Accidentally deleted grate_1FBA_allRxns again. Copied saved variable …

fff7c00

…into local repo.

Last day of SURE. Wrote code to check if all flux for a reaction were…

eddf8a9

… positive or negative.

Changed name of the 1 subfolder in VariablesSaved

17dbdea

Wed 7/24

970a15d

Pulled PC updates to desktop & resolved conflicts.

742c1ab

Sep 9

6b6a6bf

Merge branch 'egem2' of https://github.com/sriram-lab/egem into egem2

b7db698

ScottCampit reviewed Sep 26, 2019

View reviewed changes

ScottCampit requested changes Sep 26, 2019

View reviewed changes

ScottCampit assigned ScottCampit and fanela Sep 26, 2019

ScottCampit added bug Something isn't working enhancement New feature or request labels Sep 26, 2019

ScottCampit force-pushed the master branch from 52d712f to da8992e Compare October 7, 2020 20:40

		@@ -0,0 +1,419 @@
		%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%



		[ix, aapos] = ismember( mediareactions1(20:37),acetylation_model.rxns);
		posgluc = 1385; % glucose uptake reaction in recon1.



		model3.c(rxnpos) = epsilon_acetylation;
		model3.c(rxnpos) = epsilon_methylation;

Conversation

fanela commented Sep 25, 2019

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ScottCampit left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ScottCampit Oct 9, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

ScottCampit Oct 9, 2019 •

edited

Loading