Are your data normally distributed? What is your sample size and repetition? You might want to look into  generalised linear modelling (GLM), which allows you to select appropriate response functions. The free statistical software R (https://www.r-project.org) will have all you need and examples, books etc

Thanks. Well I have 6 samples in each group and data are normally distributed. 

You have two choice:

1) One way ANOVA   with 6 treatment (groups)  and 6 replication ( samples)  with 25 degrees of freedom.

2) Paired /Student t test  ( Group 1 VS  Group 2 ; 1- 3 ; 1-4 .....likewise 15 sets of comparisons. 

Reply us how you have proceeded further. Goodluck.

Thank you Professor Udikeri.

Reviewers suggestion is good.

You can simply split your 6 groups to several more appropriate statistical models with 4 groups each, that can be analyzed by ANOVA and post hoc tests.

Example

1. 1 to 4 (one way ANOVA can be also applied with Tukey post hoc)

2. 1, 2, 5, 6. (one way ANOVA can be also applied with Tukey post hoc)

3. 1, 2, 4, 6 (one way ANOVA can be also applied with Tukey post hoc)

For ANOVA analyses you also need to check for homogeneity of variance; that's actually more important than normality. 

I'd use a GLM.  You have three factors: LPS, Celecoxib, and Risperidone.  Your model should include five terms, LPS, Celecoxib, Risperidone, LPS x Celecoxib, and LPS x Risperidone.  Since you didn't have a group which got both C and R or one that got all three treatments you can't just use a simple 3-way fully crossed ANOVA.  You will then get the main effect of each of the three factors and the interactive effects of the two combined treatments.  If the interactive terms are significant, this tells you that the effects of the two treatments are not simply additive (they could be more or less than additive).  Since all the factors only have 2 levels you won't need any post-hoc tests.