Paracel TranscriptAssembler

1

• Paracel TranscriptAssembler
• Paracel GenomeAssembler
• Pavel Morozov
• 9/17/2002

2
Paracel TranscriptAssembler (PTA 2.6.2) Overview

• Key features and advantages
• Architecture
• Using PTA
• Scientific problems for PTA

3
Paracel TranscriptAssembler (PTA 2.6.2)
features.

• High capacity solution for EST-based transcript
  reconstruction, multi-processor environment.
• Complete pipeline for sequence
  cleaning,clustering, assembly
• Removal of chimeric sequences.
• Detection, alignment and visualization of
  alternative splice forms.
• Visualization through intuitive graphical
  interfaces

4
PTA advantages

• EST specific assembly engine
• Reduce number of falsely joined huge clusters by
  data cleaning and chimera detection
• PTA specifically tuned for detecting and
  alignment on of alternative splice forms
• Multi-processor capabilities
• Integrated XML format

5
TranscriptAssembler Architecture
Input ESTs
Consensus Sequences
Convert
Clean up
Clustering
Assembly
Seeded Clustering
Assembled Clusters
TranscriptView
6
Conversion of formats

• Directories or files containing sequences in
  these formats
• phd files
• Fasta
• Genbank
• EMBL
• CAML (XML)
• Obtains information about clone name,
  orientation, tissue type, etc.

7
Clean up

• Quality of the clustering and assembly results
  depends on the quality of filtering and masking
• Removes contaminants e.g. E.coil vector
• Removes low quality sequences
• Non-destructive masking on repeats, polyA tails
  and low complexity regions to keep such regions
  in final assembly.

8
Seed clustering

• Optional tool if full length mRNA is available
• ESTs with similarity above a threshold are
  clustered with a seed
• Three strategies
• Assemble only novel gene (default)
• Division of labor
• Assemble only the genes of interest

9
Clustering

• All verse all comparison
• Utilizes Haste for fast blast/fasta like
  comparison
• Uses loose clustering philosophy
• Chimera detection algorithm

10
Assembly in PTA

• Highly optimized version for EST assembly
• Utilizes quality values (supports mixed input)
• Splice variants divided into individual
  contigs/singlets
• Detects of chimeric sequences and bad ends

11
Splice Variant detection

• Alternate transcript constructs generated during
  assembly are exhaustively compared
• Determination of multiple genes per cluster
• Relative exon layout displayed in SpliceView

12
Splice Variant Alignment View
13
Output Files

• Cleanup statistics PROJECT.DATA/.scylla.stat
• Clustering statistics PROJECT.clusters.stat
• Singlets from pairwise clustering
• PROJECT.singlets.caml
• Clusters that sequences went into
• PROJECT.clusters.info
• FastA contigs and singlets from Assembly
• PROJECT.clusters.contigs
• PROJECT.clusters.singlets

14
Output Files cont.

• Assembled CAML files are stored in PROJECT.CL/
• Numbered subdirectories (100 clusters each)
• .ace files for use with consed
• .assem.caml.gz files for use with AssemblyView
• View assemblies with AssemblyView
• Cross platform Java application
• View gene transcripts and alternate splice forms
• Edit bases
• View chromatograms
• Export consensus sequences

15

• Clustering statistics
• --------------------------------------------------
  --------
• --- Total no. of sequences from prev. run
  0
• --- Total no. of current input sequences
  3955807
• --- Total no. of sequences compared
  907431
• --- Total no. of singlets after pairwise
  compared 324332
• --- Total no. of problem sequences
  8203
• --- Total no. sequences in clusters
  574896
• --- Total no. of seed clusters
  20278
• --- Total no. of clusters
  89402
• 02-08-2002 054414 PTA End.

16
Splice Variant Alignment View
17
Contig View
18
Transcript View
19
Improving Results

• Improve Cleanup parameters
• Avoid over-clustering due to false-positive
• trim appropriate sequencing artifacts
• Choose appropriate search algorithm, thresholds,
  and matrix stringency
• Rerun clusters that timed out
• Recluster clusters that are two large with higher
  threshold ? Perform multiple iterations

20
Scientific problems for PTA

• Proteomics
• Gene discovery
• Verify gene predictions for genome assembly
• Detecting splice variants
• Patterns of expression, tissue specificity
• SNP detection
• Combinations of all the above...

21
Paracel GenomeAssembler?

• All-in-one pipeline
• Base calling with TraceTuner
• Filtering and masking
• assembly
•  Easy to use graphical interface and editing
  tools
• Genomic Assemblies
• Constraints
• troubleshooting
• finishing

22
PGA Advantages

• Genome specific assembly engine
• Longer assemblies
• Fewer miss-assemblies
• Most accurate consensus calls
• Use of clone pair constraints
• Resolve repeats
• Produce scaffolds
• Parallel processing for SMP systems
• User friendly graphic interface

23
PGA architecture.
Input Files
Vector and E. coli Databases
Parameters
Base Calling w. Quality Values
Vector and Contaminant Screening
Read-pair Constraints Generation
Assembly
User interaction Processing step
Screening results PFPView
ScaffoldView ContigView
Output Files
24
Basecalling in PGA - Optional

• Using Paracels TraceTuner? technology
• advanced peak processing ( important at the ends
  of reads)
• More accurate quality value assignments
• Dye blob processing
• Heterozygote detection
• Calibration for sequencers
• ABI 377 DNA sequencer (DP, DT)
• ABI 3100 genetic analyzer
  (Pop6-BD1,2)
• ABI 3700 DNA analyzer (Pop5-BD1,2,
  Pop6-BD1,4)
• Standard calibrations used for other
  major sequencers

25
TraceTuner Heterozygote detection.

• Calibrations for
• ABI 377 DNA sequencer (DP, DT)
• ABI 3100 genetic analyzer (Pop6-BD1,2)
• ABI 3700 DNA analyzer (Pop5-BD1,2, Pop6-BD1,4)
• Standard calibrations used for other major
  sequencers

26
TraceTuner Dye Blob processing
27

• Clean up Stage
• Annotates low complexity with dust
• Masks Vector contaminants
• Filters sequences with E. coli contamination
• Constraint Generation
• Clone pair information
• Maximum clone length boundaries
• mindist
• Minimum clone length boundaries
• maxdist

28
Double-ended reads resolve repeats

• A key feature in PGA forward-reverse clone
  constraints
• Models double-ended sequencing reads
• One read is anchored in unique sequence
• Distinguishes which repeat instance the other
  read lies in

Repeat
Repeat

• Assembly WITHforward-reverse constraints

Repeat

• Assembly WITHOUTforward-reverse constraints

Misassembled fragment
leaves a singleton
29
PGA generates scaffolds
Contigs
Scaffold of ordered, oriented contigs

• PGAs forward-reverse constraints help generate a
  scaffold
• Contigs are oriented and ordered
• Gaps are bounded and may be relatively small (or
  closed)
• Much more informative result than unordered
  contigs, useful in finishing and for low-pass
  sequencing

30
Running PGA

• PGA Launcher
• Interactive Mode
• Command Line

31
PGA Launcher
32
Interactive Mode

• Enter the Command pga
• Displays introductory message and instructions
• Answer questions
• Project name
• Data location
• Constraint information
• Filtering
• Processors

33
Command line

• Advanced
• Recommend running default first
• pga ltinput datagt ltoptionsgt
• Command line options pga h (Chapter 4)
• Param file prm ltfile namegt (Chapter 9)
• Pga v gives version

34
Paracel GenomeAssembler? produces

• Fasta file of Contig and singlet sequences
• Statistical files contig sizes, number and
  identity
• Statistical information for cleanup stage
• Linking information between contigs used to
  generate scaffolds

35
End of PGA log file
36
PGA Lancher View
37
Constraints View
38
Contig View
39
Scaffold View