isnan problem in Matlab

[CicoZhang]

I am not a Matlab user, so when I run the code for a try, I am just confronted with some errors; could you please give me some help?

[AP,AP_tab,K,cost,mincost]=FitActivities(a,data_red,kin,1,1000) Error using ComputeCostGlobal (line 25) Undefined function 'isnan' for input arguments of type 'cell'.

Error in FitActivities>@(ts)(ComputeCostGlobal(ts,tk,a,data_red,ncon)) (line 36) [ts,J1]=fmincon(@(ts)(ComputeCostGlobal(ts,tk,a,data_red,ncon)), ts, [], [], [], [], lb1, ub1, [], options);

Error in fmincon (line 543) [initVals.f,initVals.g] = feval(funfcn{3},X,varargin{:});

Error in FitActivities (line 30) parfor i=1:iter

Caused by: Failure in initial user-supplied objective function evaluation. FMINCON cannot continue.

When I reproduce it following the tutorial, the problem still persists. I am using Matlab 2015b

[marcel-mischnik]

Hi Cico,

could you maybe send me your data_red variable as .mat file? I will then try to reproduce your error.

Best regards, Marcel

2016-01-21 23:01 GMT+01:00 Cico notifications@github.com:

I am not a Matlab user, so when I run the code for a try, I am just confronted with some errors; could you please give me some help?

[AP,AP_tab,K,cost,mincost]=FitActivities(a,data_red,kin,1,1000) Error using ComputeCostGlobal (line 25) Undefined function 'isnan' for input arguments of type 'cell'.

Error in FitActivities>@(ts)(ComputeCostGlobal(ts,tk,a,data_red,ncon)) (line 36) [ts,J1]=fmincon(@(ts)(ComputeCostGlobal(ts,tk,a,data_red,ncon)), ts, [], [], [], [], lb1, ub1, [], options);

Error in fmincon (line 543) [initVals.f,initVals.g] = feval(funfcn{3},X,varargin{:});

Error in FitActivities (line 30) parfor i=1:iter

Caused by: Failure in initial user-supplied objective function evaluation. FMINCON cannot continue.

— Reply to this email directly or view it on GitHub https://github.com/marcel-mischnik/IKAP/issues/1.

[CicoZhang]

Hey, Marcel: I have attached the data here. Actually, I just reproduce the method using the data in the paper and making the phosphositePlus data myself, the problem still persists.

data_red.zip

[marcel-mischnik]

The mitosis raw data in the supplement on the Bioinformatics homepage include additional columns that should be removed before starting the algorithm. I have attached the readily prepared data-mit, as well as the data_red variable. You can also use the files in the supplement archive on github. Anyway, you should fit over all conditions in order to gain better estimates for the affinity parameters, and even though 1000 iterations will give you a good coverage of the parameter space, it will probably take way too long. So, if you just want to test the method you should write something like:

[AP,AP_tab,K,cost,mincost]=FitActivities(a,data_red,kin,6,5);

2016-01-22 12:39 GMT+01:00 Cico notifications@github.com:

Hey, Marcel: I have attached the data here. Actually, I just reproduce the method using the data in the paper and making the phosphositePlus data myself, the problem still persists.

data_red.zip https://github.com/marcel-mischnik/IKAP/files/100695/data_red.zip

— Reply to this email directly or view it on GitHub https://github.com/marcel-mischnik/IKAP/issues/1#issuecomment-173893238.

[CicoZhang]

Hi, Marcel: Thank you for your help, I have pinpointed the problem. It works now. Actually, I have a another question: What do these data values (negative, positive or NaN) stand for (they are measured values, are they)?

In the formula below, K and p will be evaluated, will they? If so, in the paper (as follows), 'a' is mentioned to be 'mostly known by proteomics data'. I cannot make out why 'a' is mentioned if K is evaluated and not calculated based on 'a'?

What is the measured value 'P'?

[marcel-mischnik]

Hi Cico, the values correspond to the log2 intensity of the phosphopeptide normalized by the protein abundance. The columns stand for the six cell cycle stages G1, G1/S, early S, late S, G2 and M.

2016-01-23 14:04 GMT+01:00 Cico notifications@github.com:

Hi, Marcel: Thank you for your help, I have pinpointed the problem. It works now. Actually, I have a another question: What do these data values stand for? The intensity or counts of the peptides in different conditions (here are 6 conditions)? [image: image] https://cloud.githubusercontent.com/assets/11710225/12530280/0bb0e396-c1da-11e5-82df-4c819e80af13.png

— Reply to this email directly or view it on GitHub https://github.com/marcel-mischnik/IKAP/issues/1#issuecomment-174185068.

[CicoZhang]

Hi, Marcel: Thank you for your reply. You mean the log2 intensity of phosphopeptide over the protein abundance the phosphopeptide belongs to, is it right?

By the way, in the matlab codes, the formula is not used directly, since k is trained directly in the grid search, is it?

Best wishes,

Cico

[marcel-mischnik]

Yes and yes :).

2016-02-02 16:50 GMT+01:00 Cico notifications@github.com:

Hi, Marcel: Thank you for your reply. You mean the log2 intensity of phosphopeptide over the protein abundance the phosphopeptide belongs to, is it right?

By the way, in the matlab codes, the formula is not used directly, since k is trained directly in the grid search, is it?

https://cloud.githubusercontent.com/assets/11710225/12530319/0181f688-c1dc-11e5-81a4-9a5015f15230.png

Best wishes,

Cico

— Reply to this email directly or view it on GitHub https://github.com/marcel-mischnik/IKAP/issues/1#issuecomment-178649932.

[CicoZhang]

Thanks and thanks. :+1: