Genome BioInformatics Research Lab

  IMIM * UPF * CRG * GRIB HOME SOFTWARE * gff2aplot * HowTo * Splice Sites
 
Comparative Analysis of Orthologous Splice Sites
 
   

Summary

We will apply colors to different alignment features to visualize relationships among splice sites and the current gene annotation for an orthologous sequence pair between human and mouse. You will find this useful when integrating different data into a single figure. This tutorial also shows how to apply score-dependent color gradients to alignment features. You will learn how to project axes annotation features into the alignment panels, for instance, to pinpoint the real splice junctions from the rest. Current version of gff2aplot allows to emulate color blending to enhance the visualization of those projections.

NOTE.- For the sake of clarity, we are going to use long names for the comand-line switches. See the command-line help if you prefer short names for those cases in which a short name is available.
Bitmaps for the examples were generated as PNGs (Portable Network Graphics). If your browser is not ready for such format yet, you can visualize the PDF or PS versions by clicking on the links below each snapshot. Links to customization files, log files, GFF input files and output PostScript figures, are also available on each command-line shown.

Contents


Visualizing Orthologous Splice Sites


All subsequences matching GT (splicing donors) and AG (splicing acceptors) were extracted from the KCRB locus human and mouse orthologous sequences. A gene prediction tool was used to score their likelyhood to be real splice sites, based on position weight matrices specific for donor or acceptor sites. After that, homologous sites were encoded into APLOT pseudo-GFF format, saved as "KCRB.GT.hs-mm.gff" and "KCRB.AG.hs-mm.gff".

Donors Plot
 [PNG] [PS] [PDF
   
gff2aplot.pl              \
    --verbose             \
    --title "DONOR SITES" \
    -- KCRB.hs.gff        \
       KCRB.mm.gff        \
       KCRB.GT.hs-mm.gff  \
     > KCRB.GT.hs-mm.ps   \
    2> KCRB.GT.hs-mm.log 
  

Above figure shows the output of gff2aplot when it was run with the human/mouse donor sites analysis. Below one was made for the human/mouse acceptor sites. Both were performed under program default settings.

Acceptors Plot
 [PNG] [PS] [PDF
   
gff2aplot.pl                 \
    --verbose                \
    --title "ACCEPTOR SITES" \
    -- KCRB.hs.gff           \
       KCRB.mm.gff           \
       KCRB.AG.hs-mm.gff     \
     > KCRB.AG.hs-mm.ps      \
    2> KCRB.AG.hs-mm.log 
  

Merging Alignment Plots


In this splice sites analysis could be interesting to plot together donors and acceptors. It is as easy as putting both human and mouse files into a single command-line, as follows:

Merging Donors and Acceptors
 [PNG] [PS] [PDF
   
gff2aplot.pl                                \
    --verbose                               \
    --title "DONOR + ACCEPTOR SPLICE SITES" \
    -- KCRB.hs.gff                          \
       KCRB.mm.gff                          \
       KCRB.GT.hs-mm.gff                    \
       KCRB.AG.hs-mm.gff                    \
     > KCRB.GT-AG.hs-mm.ps                  \
    2> KCRB.GT-AG.hs-mm.log 
  

The main drawback of this simple approach is that we cannot distinguish between donors and acceptors in the alignment panel.

Projecting Axes Annotation


In order to facilitate the analysis of the merged splice sites data set, we assign different colors to those features. We have named donor features as "alignGT" and acceptor ones as "alignAG", then we set the GFF-features variable "alignment_color" to darkred and darkblue respectively. We switch on the variable "alignment_scale_color" too, so feature score will define the color, the intensity of that color being directly proportional to the score.
We also want to take into account the known gene structures. It would be better if we could have a reference to the annotation coordinates. That is achieved by projecting such annotation features into the alignment panels. The variables controlling how to represent those projections are "show_ribbons", "ribbon_style" and "ribbon_color". As you can see on the customization file (take a look to "GT-AG.rc" file), we must switch on the projections, from now on "ribbons", and then set their color and/or style (if you like). We have choosen lightyellow and palegrey to fill exonic and UTR ribbons respectively.

Customizing
 [PNG] [PS] [PDF
   
$BIN/gff2aplot.pl                                        \
         --verbose                                       \
         --title "DONOR + ACCEPTOR SPLICE SITES"         \
         --subtitle                                      \
           "DONORs (GT) in RED - ACCEPTORs (AG) in BLUE" \
         --x-sequence-coords 0..4200                     \
         --y-sequence-coords 0..4600                     \
         --custom-filename GT-AG.rc                      \
         -- KCRB.hs.gff                                  \
            KCRB.mm.gff                                  \
            KCRB.GT.hs-mm.gff                            \
            KCRB.AG.hs-mm.gff                            \
          > KCRB.GT-AG.hs-mm.color.ps                    \
         2> KCRB.GT-AG.hs-mm.color.log 
  

You may agree that the current version is much more informative than any other previous one.

Blending Projection Colors


The lastest plot is almost perfect altought the horizontal "ribbons" overlap all the vertical ones. PostScript does not allow transparent layers, but we can emulate then in a different ways. We have choosen complementary color-fill patterns to resolve the issue of the overlapping "ribbons". The parameters to enable this gff2aplot feature were collected into a new customization file: GT-AG_blend.rc. The results can be seen below:

Color Pseudo-Blending
 [PNG] [PS] [PDF
   
gff2aplot.pl                                        \
    --verbose                                       \
    --title "DONOR + ACCEPTOR SPLICE SITES (II)"    \
    --subtitle                                      \
      "DONORs (GT) in RED - ACCEPTORs (AG) in BLUE" \
    --x-sequence-coords 0..4200                     \
    --y-sequence-coords 0..4600                     \
    --custom-filename GT-AG.rc                      \
    --custom-filename GT-AG_blend.rc                \
    -- KCRB.hs.gff                                  \
       KCRB.mm.gff                                  \
       KCRB.GT.hs-mm.gff                            \
       KCRB.AG.hs-mm.gff                            \
     > KCRB.GT-AG.hs-mm.color_blend.ps              \
    2> KCRB.GT-AG.hs-mm.color_blend.log 
  

The complementary pattern approach is device dependant, it only works on those devices implementing PostScript Level 2 or higher (current PostScript interpreter versions are releases of Level 3). It is also scale dependant, this means if you plan to reduce your plot, it is better to modify the "color_merge_factor". This variable allows numerical values between 0.07 and 2. That is what we have done on the new customization file provided, GT-AG_sm.rc, increasing value from default 0.5 to 1.

Scaling Down Color Blend
 [PNG] [PS] [PDF
   
gff2aplot.pl                                        \
    --verbose                                       \
    --title "DONOR + ACCEPTOR SPLICE SITES (III)"   \
    --subtitle                                      \
      "DONORs (GT) in RED - ACCEPTORs (AG) in BLUE" \
    --x-sequence-coords 0..4200                     \
    --y-sequence-coords 0..4600                     \
    --custom-filename GT-AG.rc                      \
    --custom-filename GT-AG_sm.rc                   \
    -- KCRB.hs.gff                                  \
       KCRB.mm.gff                                  \
       KCRB.GT.hs-mm.gff                            \
       KCRB.AG.hs-mm.gff                            \
     > KCRB.GT-AG.hs-mm.color_small.ps              \
    2> KCRB.GT-AG.hs-mm.color_small.log 
  

You can observe that now the "ribbons" look like pixelized. If you scale down twice the last figure, the overlapping regions for the annotation projections will print out like in the previous one (or pretty similar).

 
  Disclaimer webmaster