Supported by the NSF Plant Genome Research and REU Programs

1
Tutorial of bioinformatics and tree generation at
the Cell Wall Genomics website
Bryan Penning
Supported by the NSF Plant Genome Research and
REU Programs
2
Bioinformatics Goals

• We currently have a wealth of Arabidopsis
  thaliana cell wall gene information on the
  website, we wanted to
• Add family information about rice and maize Type
  II cell walls to compare to A. thaliana Type I
  cell walls
• Add links to outside information on rice genes
  like we have for A. thaliana
• Include annotated composite trees of A. thaliana,
  rice and maize gene families
• Add links to sites used to generate the data
• Add source protein sequence used for our family
  trees so other researchers can make their own
  adding their genes of interest
• Generate a tutorial on how researchers can make
  use of the bioinformatics data on our site

3
Diagram of our bioinformatics approach
Too few genes, Blast other sites
N
Blast TIGR
Choose genes
Make tree
Y
Too many genes, tighten criteria
N
Diagram of the process used to find the genes and
draw family trees for cell wall related rice
genes. The same approach is used for maize.
4
Diagram of our bioinformatics approach
A. thaliana genes
Draw tree with all family members
Publish to web
Annotate
Rice genes
Diagram of the process used to integrate cell
wall related genes from all three family trees
into a composite tree.
Maize genes
5
BLASTing genes

• To be considerate of the bioinformatics community
  with the number of BLASTs to be performed and to
  speed the process, we downloaded the text or
  flat file of the TIGR rice protein sequences
  (available at http//www.tigr.org/tdb/e2k1/osa1/d
  ata_download.shtml) and performed local blasts
  using blastall from NCBI (available at
  http//www.ncbi.nlm.nih.gov/BLAST/download.shtml)
• Direction for use of these tools is available at
  the above sites and is beyond the scope of this
  tutorial
• For a small number of BLASTs, you can use
  web-based methods and common programs such as
  Word and Excel plus any of a number of
  downloadable tree drawing programs to make these
  kinds of trees on your own if you are not
  familiar with programming languages such as Perl
  to automate the process. Although web searches
  can be more time consuming, they work just as
  well for a few sequences

6
Web BLASTing

• For smaller numbers of BLASTs to the rice genome,
  TIGR provides an excellent Web BLAST at
  http//tigrblast.tigr.org/euk-blast/index.cgi?proj
  ectosa1
• You can also use the new BLAST tool at Gramene
  http//www.gramene.org/multi/blastview for most
  cereal sequences
• Note gene model versions sometimes differ
  between Gramene and TIGR as one site may update
  to the latest model before the other

7
Web BLASTing

• Downloading the protein sequence for Arabidopsis
  SUD1 (At3g46440) from TIGR, you can BLAST it
  against the TIGR Rice Pseudomolecules Protein
  database using BLASTp

8
Web BLASTing

• You get a series of hits to the gene of
  interest
• A higher score and smaller probability is a
  better match to the original gene
• This procedure is followed for all of the genes
  in a family to gather the best possible hits,
  sort the hits to remove duplicates and choose the
  best rice matches to the Arabidopsis families
• You can use NCBIs blastall tool for multiple
  simultaneous blasts as we do for this step

9
Organizing BLASTs

• This is a word document generated by BLASTing
  SUD1 and SUD2 of Arabidopsis against the TIGR
  Rice Protein database
• The hits were copied into word and set to the
  font Courier New, 9 pt and saved as a text only
  document (to remove the HTML code)
• The file was reloaded in Word and converted to a
  table (table menu) using other and the character
  (shift \) to separate the columns

10
Organizing BLASTs

• The Word file is copied into Excel and the Data
  Sort menu is used to sort by the first column
• This brings all of the same named genes together
  (the two highlighted lines for example)
• Duplicate genes are removed from the spreadsheet
  and the far right column only (LOC_Osxxgxxxxxx)
  tags can be copied back to word

11
Organizing BLASTs

• You can use the table menu to convert table to
  text (Paragraph Marks) to generate a list of
  genes
• These genes can be searched through a downloaded
  database using the NCBI fastacmd (included in the
  BLAST download tools) or you can search them one
  at a time using a web-based database such as the
  locus search name on TIGR (http//www.tigr.org/td
  b/e2k1/osa1/LocusNameSearch.shtml)

12
Generating a tree

• Once you have found all of your sequences, check
  that each sequence name has a (denoting a new sequence name) and the sequence
  starts on a new line
• Copy and paste all of your sequences into an
  alignment program like ClustalW (we use
  http//align.genome.jp/ from the Kyoto University
  Bioinformatics center, but any ClustalW program
  will work)

13
Generating a tree

• For our trees we use Slow/Accurate pair-wise
  comparisons and Gonnet for our Weight Matrix (two
  spots on the website)
• Click execute alignment to get your sequence
  alignment
• At the end of the alignment page will be the
  information needed for tree drawing programs
• You can click on clustal.dnd for a quick tree or
  take the information after it A Newick format
  tree and copy it into a new Word file, saving it
  as a text file (include all parenthesis)

14
Creating a tree

• We use the program TreeDyn to generate our trees
  (available at http//www.treedyn.org/)
• This is an example of the Arabidopsis and rice
  1.1 family
• The tree text file was loaded into TreeDyn and
  the frame enlarged
• The red text for Arabidopsis sequences was done
  by changing the font color to red and using the
  find panel to find all At sequences (which turn
  red)
• The scale at the bottom was added by right
  clicking on that space and choosing the tree
  name, annotation, and scale sub-menus
• This square tree is useful to see associations of
  genes for different species

15
Square tree example

• This is part of the family 1.1 square dendrogram
  of Arabidopsis, rice and maize from our website
• The red names are Arabidopsis sequence, the black
  names are rice, and the green names are maize
• Regions alternate between grey shaded and white
  backgrounds (added with Photoshop) to indicate
  clades of similar sequence genes which may relate
  function (such as AUD/SUD or GME, etc)

16
Radial dendrograms

• TreeDyn can also draw radial dendrograms such as
  the one shown for rice family 1.1
• This can be done by right clicking on the tree
  area to bring up the grey box in TreeDyn,
  choosing your tree, then Conformation- Radial
• Treedyn allows you to resize, rotate, and flip
  clades around (see http//www.treedyn.org/ for
  detailed tutorials on these processes)
• For our site, we export the radial trees as jpeg
  images

17
Finishing a radial dendrogram
The TreeDyn tree jpeg is finished as a FLASH file
where the ovals and family names are added (Rice
family 1.1 shown)
Each individual clade of a family tree is also
prepared in TreeDyn and link buttons added later
in FLASH (AUD/SUD-like shown)
18
Viewing your gene of interest

• We provide protein sequence information you can
  download and add in your own sequence of interest
  for comparison to these three species
• Under each tree (family 1.1 shown) is the link
  View the protein sequence file
• Right click and choose Save Target as to
  download the sequence with a filename and
  location you will remember
• You can do this for each Arabidopsis, rice, and
  maize family

19
Viewing your gene of interest

• You may have a sequence you think is related to
  a particular family such as nucleotide
  interconversion pathway (family 1.1)
• For example, the wheat EST CV523101 from
  Genebank
• http//www.ncbi.nlm.nih.gov/entrez/viewer.fcgi?db
  nucleotidevalCV523101
• might be related to the TIGR rice gene
  Os05g29990 in the AUD/SUD clade of family 1.1
  according to information from Gramene

20
Viewing your gene of interest

• You can take the nucleotide sequence and covert
  it to protein sequence using a program such as
  Genemark (http//opal.biology.gatech.edu/GeneMark
  /eukhmm.cgi)
• Protein sequence returned
• CV523101_wheat
• IARIFNTYGPRMCIDDGRVVSNFVAQALRKEPLTVYGDGKQTRSFQYVS
  DLVEGLMRLMEGDHIGPFNLGNPGEFTMLELAKVVQDTIDPNARIEFREN
  TQDDPHKRKPDITKAKEQLGWEPKIALRDGLPLMVTDFRKRIFGDQDSAA
  TATEG

21
Viewing your gene of interest

• Paste all of the sequences for family 1.1
  (Arabidopsis, rice, and maize) plus the Wheat
  EST, CV523101_wheat, converted to protein into a
  ClustalW program such as
• http//align.genome.jp/
• from the Kyoto University Bioinformatics center
• Perform the multiple alignment, copy the Newick
  tree data generated into a new word file, and
  save a text file as previously shown

22
Viewing your gene of interest

• Taking the Newick tree from clustalW into TreeDyn
  as previously shown will allow you to visualize
  the tree
• The AUD/SUD clade of the tree generated by
  TreeDyn shows that the wheat EST (in blue) is
  most closely related to the rice gene Os05g29990
  in the AUD clade

The AUD/SUD clade of the family 1.1 tree for
Arabidopsis (red), Rice (black), Maize (green),
and a wheat EST (blue) added to demonstrate how
you can visualize relatedness of your own genes
using our protein sequences
23
Bioinformatics sites used

• General
• Multiple alignment for trees, ClustalW
  (http//align.genome.jp/)
• Making trees, TreeDyn (http//www.treedyn.org/)
• BLASTing NCBI (http//www.ncbi.nlm.nih.gov/BLAST/)
  
• Proteins translated by GeneMark
  (http//opal.biology.gatech.edu/GeneMark/eukhmm.cg
  i)
• Rice
• Sequence BLAST using TIGR (http//www.tigr.org/tdb
  /e2k1/osa1/)
• Downloading rice protein sequences from TIGR
  (http//www.tigr.org/tdb/e2k1/osa1/LocusNameSearch
  .shtml)
• Maize
• Sequence BLAST using TIGR ZmGI (http//www.tigr.or
  g/tigr-scripts/tgi/T_index.cgi?speciesmaize)
• Sequence BLAST using Gramene (http//www.gramene.o
  rg/multi/blastview)