I'm aware of density plots around TSS. These are very straight forward. However I am trying to to do the same, by using alternative exons instead of TSS/promoters. It seems very tricky, especially because of variable length of exons. Does any one know how to plot these profiles ?
Isn't it just like average gene analysis, where the length of the feature doesn't matter because you divide each feature into the same number of bins so that data across a large number of features can be displayed as an average conglomerate? It's like taking column averages of a large table representing the bins of the features (assuming all features have been divided into say 20 bins, regardless of feature size), and plotting that. Have you read the materials and methods of the referenced paper? Don't they say there? (you might mention the author of the paper, or a better way to look it up instead of a deep link to one figure).
I think it is not like "average gene analysis". Because the intron length is constant but not the exon. Yes there is a brief description available regarding this plotting, i.e. "normalized complexity" which will highlight common features shared by multiple transcripts.
My understanding of such graphs and really the data behind them is the different exon lengths are standardized to a common value while the intronic features are displayed as measured from the respective end of the exon. In other words, all exons (length = n) treated as equal in size and the measures taken within the exon are measured in actual base pairs from either the 5' or 3' end of the respective exon.
So, make all exons oriented from left to right, 5' to 3'[?]
All features to the left are measured from base 1 of the 5'end[?]
All features to the right are measured from base n of the 3' end[?]
Thus, an "upstream" feature is never measured relative to the 3' end and vice versa.
I think it is not like "average gene analysis". Because the intron length is constant but not the exon. Yes there is a brief description available regarding this plotting, i.e. "normalized complexity" which will highlight common features shared by multiple transcripts.