urgent help required - data analysis of time course

#1
Hi all
Could someone please help me with analysis of below data?
I am trying to compare gene expression in two different virus strains in a time course. My last time point looks significant but based on Prism's Sidak test twoway anova it comes non-significant.
Could someone please help?

data are transformed (nonlogarithmic). comparison of vaccine vs virus (3 replicate each).

Columns: vaccine1, vaccine2, vaccine3, virus1, virus2, virus3 .

rows: hours post infection: 6, 12, 18, 30, 42, 54, 66, 78, 90.

Data are attached.

Many thanks for your help
Yashar
 

Attachments

#2
Typically, visitors to this side come with specific questions. They do not say: "Do the whole analysis for me. Please, please." You would need to state what you have done already, what the obstacles were and which research / programming / other questions you have now... Somebody nice here responds and you go back, and you do your analysis.

I know you have started with saying what outcomes have been generated already. All of us here would appreciate if you could elaborate on that. Thanks.
 
#3
Hi staassis. Thanks for your reply. I use graphpad prism. and based on graphpad prism output, the data is not significant. I have attached the results. However, a t-test, for each time point indicate that the last time point is significant. My question is, can i use t-test to report significance of the data? is t-test relevant? Or does it need to be rm two way anova specifically? considering that the anova test does not detect any interaction between the column and row factors, is it still a valid test for this analysis?
cheers.
ys
 

Attachments

#4
Hi staassis. Thanks for your reply. I use graphpad prism. and based on graphpad prism output, the data is not significant. I have attached the results. However, a t-test, for each time point indicate that the last time point is significant. My question is, can i use t-test to report significance of the data? is t-test relevant? Or does it need to be rm two way anova specifically? considering that the anova test does not detect any interaction between the column and row factors, is it still a valid test for this analysis?
cheers.
ys
 

Attachments

#5
Need to see the research questions, description of the data + experiment, attempted statistical methods, summary of the results. Just like I mentioned in my previous post... On the other hand, you keep repeating: "I did this little thing. Here's the output.."
 
#6
Thanks staassis for your reply. And apologies for being ambigous. It was not intentional.

I am comparing two groups of viruses: vaccine strain and wildtype strain and I am measuring expression level of a RNA transcript from timepoint 0 to 90 hours post infection.
All I want to know is: If the level of expression of my RNA remains the same between the two viruses, or one of the two viruses start to express higher levels at certain hours post infection.
I am expecting that my vaccine virus start to express more RNA (significantly higher) during the last three time points.

I have attempted RM two way anova, and according to it there is no significant difference.
However, t-test for each time point indicates that at 90 hours post infection, there is a significant difference in RNA expression between vaccine and wildtype viruses.

description of data:
columns: vaccine virus or wildtype virus in triplicate.
rows hours post infection. starting from 6 hours post infection and extending to 90 hours post infection.

I hope I was more clear this time.
 
#7
Running multiple t-tests would no be right here... This is a set-up for repeated measures ANOVA with two factors:

1) Time: within-subjects factor,

2) Virus Type: between-subjects factor.

You should start with the full factorial model (including the interaction of Time and Virus Type) and drop the non-significant terms one by one. When all the remaining terms are statistically significant, that is when you are ready to make inference... Regular repeated measures ANOVA (as implemented in packages like SPSS) is suitable if either you have a lot of data in each category or if the residuals are normally distributed. Otherwise you would have to program your own randomization test (your own non-parametric version of repeated measures ANOVA).
 
#8
Running multiple t-tests would no be right here... This is a set-up for repeated measures ANOVA with two factors:

You should start with the full factorial model (including the interaction of Time and Virus Type) and drop the non-significant terms one by one. When all the remaining terms are statistically significant, that is when you are ready to make inference...
Is this more of an exploratory approach? Would the p values be smaller than they should be?
 
#10
Is this more of an exploratory approach? Would the p values be smaller than they should be?
The p-value from the omnibus ANOVA test may be larger or smaller than the p-value for a t-test comparing just two particular groups. What matters is that there is only one ANOVA p-value but many t-tests (multiple testing). And none of those t-tests summarize the whole dynamics in a statistically efficient manner.

If assumptions of an ANOVA F-test or t-test are violated, then their p-values tend to be smaller than the truth (false significance).

To me, the project did not seem to be 100% exploratory analysis. Yasharsad needs formal inference as well, with testing, p-values and all that. At the end of the day he/she wants to know if the two virus types have the same dynamics or different dynamics.