Heatmap for DEG data

Question

[Azhar]

I have normalized processed micro array data for 4 samples or columns and invariable column length i want to make heat map for this data is this good idea or not and second thing i tried to make heatmap foe this data but i am not satisfied my data look like

A   B   C   D
3.1444143   3.1282364   2.7754867   2.1431285
2.1431285   3.1282364   3.4834301   3.569108
3.569108    2.7754867   2.2201648   3.1282364
3.1282364   3.4834301   2.0201576   3.1282364
3.1282364   2.2201648   3.4834301   2.7754867
2.7754867   2.0201576   2.2201648   3.4834301
3.4834301   3.4834301   2.0201576   3.4834301
2.2201648   2.2201648   -3.2991285  2.2201648
2.0201576   2.0201576   -3.2991285  2.0201576
-3.2991285  -3.2991285  -2.0686522  -3.2991285
-3.2991285  -3.2991285  -2.0686522  -3.2991285
-2.0686522  -2.0686522  -2.4884053  -2.0686522
-2.0686522  -2.0686522  2.0201576   -2.0686522
-2.4884053  -2.4884053  -3.2991285  -3.2991285
-2.0880043  -2.0880043  -3.2991285  -2.0686522
-2.4821819  -2.4821819  -2.0686522  -2.0686522
    -2.0686522  -2.0686522  -2.4884053
    -2.4884053  -2.4884053  2.0201576
    -2.0880043      -3.2991285
    -2.4821819

so i replace the missing values with 0 and used the scale rows method like this

heatmap.2(as.matrix(data),main="data",trace="none",margins=c(5,10),cexRow=0.5,Rowv=FALSE,colv=FALSE,key=TRUE,scale="row")

but the figure it produces does not look like revelent to data

i have two questions 1. can i use heat map for invarible data cloumns cz i did not seen any case like that 2. if yes then what parameters should i use

[WouterDeCoster]

I added (code) markup to your post for increased readability. You can do this by selecting the text and clicking the 101010 button. When you compose or edit a post that button is in your toolbar, see image below:

[venu]

Just a few points

• Do you have probe/transcript ids associated with these expression values? If yes, include them in constructing heatmap
• I would remove probes with missing expression values
• Are these z-scores? If not, I would calculate and use them to generate heatmap

[Azhar]

yes i have transcript IDs but those are different for each condition or coulmn A B C and D so cannot use those because these are not common they are DEG so they are different for each sample . and these values are foldchange so can i use them for heatmap or i should use respective expression values . 2. second if remove missing values then heat map would be missing some parts of some column due to missing values and invariable length and it does not look good.

[WouterDeCoster]

those are different for each condition or coulmn A B C and D so cannot use those because these are not common they are DEG so they are different for each sample

If I understood correctly, the values (first line): 3.1444143 3.1282364 2.7754867 2.1431285 have nothing to do with each other?

[Azhar]

yes you are right these are DEG genes so i want to make heatmap for them is this good idea or not

[venu]

@Azhar, I don't think this is a good idea. In a heatmap, you show the expression of one transcript in all the samples. If the values in your first row belong to different transcripts what would you understand from a heatmap. So in your table, each row should represent expression values of a transcript in all samples.

[WouterDeCoster]

That's absolutely not how heatmaps work. You need a matrix with samples and genes, not with some randomly gathered DEG FC values.

but the figure it produces does not look like revelent to data

Not surprising at all ;-)

[Azhar]

yes you are right but what i asked is that i have data after DEG in 4 different samples so offcourse it does not explain single gene in every sample instead it can show the DE genes which are different in different samples , mu question is that can i use this data to show the expression pattern of up and down regulated genes FC or expression values for each sample which is offcourse will not have equal number of genes hence so not equal lenght of coulmn ?

[WouterDeCoster]

Try that again and take some more time structuring sentences. This is very hard to read. I understand that for non-native speakers this is maybe not so easy, but you are making it impossible for us to understand what you are looking for.

(I'm also not a native English speaker.)

[Azhar]

o sorry i tried to make it clear may be in hurry i write it too long and unstructured sentence, So I said i know i need matrix for data but is it nesscaey that data should be for each gene in each sample like A B C D a b c d but since my data is DEG sorted data so logically it should have very less common genes becuse it is sorted on differential expressed genes right! So my question is what you mean by random data ... As for i understand its not random data its differential expressed data different gene in 4 samples and columns and since use cut off parameter fold change 2 there number would be different in each sample like A B C D 3.1444143 3.1282364 2.7754867 2.1431285 2.1431285 3.1282364 3.4834301 3.569108 3.569108 2.7754867 2.2201648 3.1282364 3.1282364 3.4834301 2.0201576 3.1282364 3.1282364 2.2201648 3.4834301 2.7754867 2.7754867 2.0201576 2.2201648 3.4834301 3.4834301 3.4834301 2.0201576 3.4834301 2.2201648 2.2201648 -3.2991285 2.2201648 2.0201576 2.0201576 -3.2991285 2.0201576 -3.2991285 -3.2991285 -2.0686522 -3.2991285 -3.2991285 -3.2991285 -2.0686522 -3.2991285 -2.0686522 -2.0686522 -2.4884053 -2.0686522 -2.0686522 -2.0686522 2.0201576 -2.0686522 -2.4884053 -2.4884053 -3.2991285 -3.2991285 -2.0880043 -2.0880043 -3.2991285 -2.0686522 -2.4821819 -2.4821819 -2.0686522 -2.0686522 -2.0686522 -2.0686522 -2.4884053 -2.4884053 -2.4884053 2.0201576 -2.0880043 -3.2991285 -2.4821819

like consider in sample A gene number is 150, in B 270, in C 229, D 345 so can i represent this data in the form of heatmap

[Jean-Karim Heriche]

If your conditions are in columns and probes/genes in rows, then you probably want the z-scores by columns i.e. scale = "col"