Comparison of different dynamic models for C elegans

I use something like a ROC curve, that measures the proportion of the neurons belonging to the subnetwork and the proportion of neurons not belonging to the subnetwork that the method finds, depending on the threshold:

I use the chemotaxis network I used before.

Comparison between the model with all conductances equal (as in the older post) and the model with all connections as if gap junctions, and no membrane conductance:

>> clear
>> load(‘c:\hipertec\ablationelegans\Datos\datos_celegans_paradinamica.mat’)
>> V0=zeros(279,1);
>> V0([40 41])=1;
>> M{1}=conectividad2matrizsistema(todas.A_chem,todas.A_ej,todas.GABA,1,1,1);
>> resultados{1}=matrizsistema2experimentoablacion(M{1},V0,500,.005,[-1 1],54:57);
>> M{2}=conectividad2matrizsistema(zeros(279),todas.A_chem+todas.A_ej,todas.GABA,1,1,0);
>> resultados{2}=matrizsistema2experimentoablacion(M{2},V0,500,.005,[-1 1],54:57);
>> buenas=[11:14 19 20 77 78 111:113 168 169 178 179]; % Neurons belonging to the subnetwork
>> resultados2pseudoroc(resultados{1},buenas,[40 41 54:57],’b’);
>> hold on
>> resultados2pseudoroc(resultados{2},buenas,[40 41 54:57],’r’);

 

Now I compare the old model (with all conductances equal) with different time scales:

>> resultados{1,2}=matrizsistema2experimentoablacion(M{1},V0,250,.005,[-1 1],54:57);
>> resultados{1,3}=matrizsistema2experimentoablacion(M{1},V0,100,.005,[-1 1],54:57);
>> clf
>> resultados2pseudoroc(resultados{1,1},buenas,[40 41 54:57],’b’);
>> hold on
>> resultados2pseudoroc(resultados{1,2},buenas,[40 41 54:57],’r’);
>> resultados2pseudoroc(resultados{1,3},buenas,[40 41 54:57],’k’);

Perfect overlapping, so the time scale does not make any effect in this range.

Test of the pseudoROC:

>> resultados=rand(279,1);
>> clf
>> resultados2pseudoroc(resultados,buenas,[40 41 54:57],’k’);

OK.

Now, I compare the model with equal conductances with a model tuned so that it does not saturate:

>> M{1}=conectividad2matrizsistema(todas.A_chem,todas.A_ej,todas.GABA,1,1,1);
>> resultados{1}=matrizsistema2experimentoablacion(M{1},V0,500,.005,[-1 1],54:57);
>> resultados{1}=matrizsistema2experimentoablacion(M{1},V0,500,.005,[-1 1],54:57);
>> M{2}=conectividad2matrizsistema(todas.A_chem,todas.A_ej,todas.GABA,.05,1,1.5);
>> V=matrizsistema2V_num(M{2},V0,10000,.005,[-1 1]);
>> plot(V(54:57,:)’)

>> resultados{2}=matrizsistema2experimentoablacion(M{2},V0,10000,.005,[-1 1],54:57);
>> resultados2pseudoroc(resultados{2},buenas,[40 41 54:57],’r’);
>> hold on
>> resultados2pseudoroc(resultados{1},buenas,[40 41 54:57],’b’);

The model with equal conductances works better.

Advertisement