Transcript CGPDB Presentation

Linear graphical representation of BLAST search of
Arabidopsis genome against Lettuce/Sunflower EST assemblies.
http://cgpdb.ucdavis.edu/database/est_vs_ath/tigr_vs_let_and_sun.html
Lettuce/Sunflower EST CGPDB project.
Data analysis, assembly visualization and validation.
Alexander Kozik, Brian Chan, Richard Michelmore.
Department of Vegetable Crops, University of California at Davis, CA 95616.
Over 60,000 lettuce and 40,000 sunflower ESTs from multiple libraries have been assembled
using the CAP3 program (http://genome.cs.mtu.edu/cap/cap3.html) and organized into the
Compositae Genome Project database (http://cgpdb.ucdavis.edu/). This assembly represents
about 19,000 lettuce and 12,000 sunflower unigenes. mySQL (http://www.mysql.com/) was
chosen as an efficient tool to manage the data. Custom PHP and Python programs were
developed with publicly available php_my_admin software to manipulate the data and
visualize the assemblies. To exploit the generation of the ESTs from different genotypes
representing mapping parents of lettuce and sunflower, we developed a new software to
identify possible polymorphisms. About 250 insertions/deletions (INDELs) and 2,500
substitutions (SNPs) have been discovered for lettuce and sunflower assemblies using
custom Python scripts. Wet lab experiments have confirmed the predicted polymorphism in
~90% cases. A new clustering algorithm was used to find putative COS (conserved ortholog
set) markers. About 1,200 lettuce and 500 sunflower putative COS markers have been
identified based on clustering analysis with the complete Arabidopsis genome. EST
assemblies have been analyzed for multidomain proteins, possible chimeric clones and
misassembled contigs using graph theory and our custom Graph9 program. Clusters of
multigene
families
have
been
visualized
using
PhyloGrapher
program
(http://cgpdb.ucdavis.edu/PhyloGrapher/).
Linear graphical representation
of BLAST search against the
Arabidopsis
genome.
Each
element represents a 'gene' predicted ORF (TIGR version,
September 2002). Elements are
ordered according to position on
chromosome and are web links
to corresponding entries in the
CGP database. Color intensity
indicates level of similarity
(normalized Expectation values =
-log(Exp)). Green - significant hit
to lettuce, Red - significant hit to
sunflower. Yellow - significant hit
to both. White blocks separate
the Arabidopsis chromosomes.
Image created with PyMood
(http://www.pymood.com/)
Sequence clustering: finding chimeric and multidomain ESTs
Scheme of Data Processing and SNP/INDEL Discovery Pipeline:
Two different genotypes for each genus:
(Lettuce: cv. Salinas and L. serriola)
(Sunflower: RHA801 and RHA280)
chimeric
sequence
cDNA library construction
(individual libraries for each genotype)
Sequencing
Raw Chromatograms (reads)
processing by Phred-CrossMatch
Clustering visualized by PhyloGrapher,
for details see http://www.atgc.org/
Individual CAP3 assembly for each genus:
different genotypes analyzed together
Clustering analysis by Graph9 program:
BLAST EST assembly against itself -->
Processing of the CAP3 output with
Finding in the assembly all mismatches in individual
sequences versus consensus sequence. If all mismatches
for given position belong to one genotype it is considered
as a potential polymorphic site (SNP or INDEL)
Contig Viewer
--> Generation of "Matrix" file using
tcl_blast_parser.tcl program -->
custom Python scripts and generation
of tab-delimited files ready to go into
relational mySQL database
http://cgpdb.ucdavis.edu/database/chromat_viewer/ContigViewer_MMX.php
Graph9 output with bridges info,
see table lettuce_clustering at CGPDB
http://cgpdb.ucdavis.edu/ for details
Conserved Ortholog Set (COS) Markers candidates
Pipeline to process BLAST output:
Blast parser generates "Matrix" file form regular BLAST output.
Graph9 program analyzes "Matrix" file and generates "Group Degree Info" file.
"Group Degree Info" file contains full information about sequence clustering based
on "Matrix" file. See http://cgpdb.ucdavis.edu/BlastParser/Blast_Parser.html
on-line Contig Viewer is a set of PHP
scripts to navigate assembly in
full details. Contig Viewer displays
information about assembly, highlights
sites of polymorphism, provides web
links to BLAST reports for consensus
and individual sequences.
All underlying data are stored in mySQL
database. There are four tables that
provide full information to display
assembly graphically. All tables were
derived by processing of CAP3 output by
custom Python scripts.
Table with overlap info
for every sequence in the assembly
--> Clustering and bridges search
by Graph9 program.
Example of false “single” hit
Strategy to identify COS
candidates:
Clustering analysis using Graph9
program and removing from
potential COS set all ESTArabidopsis clusters with multiple
Arabidopsis nodes. Clustering
parameters were: Expect cutoff
1e-10, Identity cutoff 20% and
Overlap cutoff 50 amino acids.
Table with CAP3 “clip” info
for every sequence
Table with mismatch info
sequences vs consensus of the assembly
CAP3 assembly output files are sufficient to extract full information about
polymorphic sites. Besides
numerical information, CGPDB
provides full access to
raw chromatograms for every
sequence in the database.
Therefore base calling can be
verified for every nucleotide
in lettuce/sunflower ESTs
Table with tissue info
for every sequence
Graphical representation of BLAST search lettuce, sunflower, tomato and corn ESTs against
Arabidopsis genome. Potential conserved orthologs. Color scheme: lettuce&sunflower - green,
tomato - red, corn - blue. Additive color mixing reflects EST representation for Arabidopsis gene
(ORF). white = red + green + blue, yellow = red + green, cyan = green + blue, purple = red + blue.
Genes are web links to corresponding entries in CGP database
(http://cgpdb.ucdavis.edu/database/est_vs_ath/arabidopsis_cos_map.html)