Hello.
Could you give me a hint how to make 2d or better 3d plots where two or more separated groups of conditions can be shown? My data set looks like this:
Gene CON1 CON2 CON3 condiditons 17 in totoal -->
A
B
C
D
E
genes 116 rows in total
Table id filled with expression intensities. How should I calculate distances between conditions based on expresion intensites? At this moment I made plots with two groups when every point was a gene(116 points) but I'd like to do this for conditions.
I've made a heatmap and it looks cool, I see clearly separated group of genes, I also made a plot using scatter3d and MDS, but they are also show groups of genes,.. The problem is with plot which shows clusters of conditions (con1 con2 con3 - close to each other, the rest in other group). Example dataset:
Gene CON1 CON2 CON3 CON4
A 0 1 124 245
B 3 4 67 82
C 2 0 112 100
D 112 231 4 18
E 92 181 4 0
F 251 124 21 3
Sorry, wasn't quite clear. You mean your control conditions are grouped relatively tight to your treatments?
How many replicates do you have per treatment and also what are you trying to answer in a biological context? If it is which genes are sig. up and down-regulated I recommend looking at the Edger OR Deseq2 vignette. I've been working on it a lot lately, trying to learn how and why it works. Can maybe provide some more feedback from my experiences.
I've made solution for my data, I used PCA and MDS methods, I've made also 3d plots for this data. Data structure is quite simple, first column Gene_ID next coumns were next conditions, I mean tissue for example, let's take 3 columns: brain, intestine, blood rows were genes, one row for example: RASA1 12 76 124 numbers are expression intensity. I can show you my scripts in R and results if you ask, Regards Adam
Log in to answer this question.
Some hints : MDS plots with R
Limma has a function plotMDS as well as CummeRbund, I have used the one from CummeRbund works well.
I have a write-up for PCA and MDS here https://rpubs.com/crazyhottommy/PCA_MDS
Great tutorial, thanks for help, I get what I need ;)