Variation Detections and De novo Assemblies from Next-gen Data Zemin Ning The Wellcome Trust Sanger Institute

Outline of the Talk:  Projects before Bioinformatics  Bioinformatics Projects Involved  Variation Detection SNP, Indel, CNVs etc  Fuzzypath – short read assembly  Extremely GC Biased Genomes

Powder Simulation

Hair Dynamics Genetics and Human Hair Structure AFRICAN CAUCASIAN EAST ASIAN

 SSAHA (Sequence Search and Alignment by the Hashing Algorithm Ssaha2 – Alignment tool for Solexa, 454, ABI capillary reads ssahaSNP – SNP/indel detection, mainly for ABI capillary reads ssahaEST – EST or cDNA alignment ssaha_SV – Structural variation (CNVs) detection ssaha_pileup – SNP/indel detection from next-gen data  Phusion Development and maintenance of the pipeline Production of WGS assemblies: Mouse, Zebrafish, Human (Venter genome), C. Briggsae, Rice, Schisto, Sea Lamprey, Gorilla, Malaria and many bacterial genomes  TraceSeach Public sequence search facility for all the traces  Fuzzypath Short read assembler Informatics Projects Involved

Read mapping by hashing and dynamic programming data base of subject sequences FASTQ file with query sequences banded Smith-Waterman alignment

Pipeline of ssaha_pileup Sequencing Reads SNP File Ssaha_cigar Alignment - ssaha2 Unique placed cigar read file SE Reference fasta Pileup/cons PE Ssaha_pairs Ssaha_clean Ssaha_indelssaha_pileup Indel File

Mapping Score in ssaha2 Read mapping score is used to assess the repetitive feature of the read in the genome. In the cigar file cigar::50 S map = 50 is the mapping score: R = read length; S max - maximum alignment score (smith-waterman) of the hits on genome; S max2 - second best alignment score of the hits on genome; Say you have one read of 30 bases which has a few hits on the genome: Best hit: exact match with S max 30; Second best hit: one base mismatch with S max2 29. The mapping score for this read is S map = 10; Read Reference

SNP score is calculated as the sum of weighted read mapping scores, combined with base quality. For Solexa reads: S map - read mapping score, from 0 (repeat) to 50 (unique); F q - base quality factor: F q = 1 if Q>=30 F q = 0.5 if Q =30 F q = 0.5 if Q<30; N – number of read coverage at the location. SNP Confidence Score in ssaha2

Getting Personal with J. Craig Venter and James Watson

Datasets n Venter: ABI capillary reads –Celera: 19,397,599 55% in pairs –JCVI: 12,541,352 98% in pairs –Total: 31,938,95172% in pairs n Watson: 454 GS FLX reads –Baylor & Roche 74,198,831 –single end reads with length 150 – 280 bps n Chromosome X Illumina reads –140 million paired Solexa reads at ~45x

IndividualsCount% dbSNP Venter SNP Calling (Capillary) Homozygous SNPs % Heterozygous SNPs % Total SNPs % Watson SNP Calling (454) Homozygous SNPs % Heterozygous SNPs % Total SNPs % X Chromosome SNPs (Solexa) Homozygous SNPs % Heterozygous SNPs % Total SNPs % SNP Results from Three Individuals

Deletio n Insertion Reference Sequence Sample Reads VNTR A’’ A’ Insertion Sample Reads Reference Sequence        1 1’ 2’ 2 Sample Reads Detection of Structural Variations

Deletion VNTRs Insertion Total number: Maximum length (bp): Minimum length (bp): Average length (bp): Affected Bases: Structural Variations against NCBI36 Deletion VNTR Insertion Total number: Maximum length (bp): Minimum length (bp): Average length (bp): Affected bases:

Deletion – Size Distribution

VNTRs – Size Distribution

Simulated Solexa reads : Number of reads: 25,647,985 Genome size: 23.0 Mbp Read length:36 Read coverage:40x Num. of uniquely placed PE reads: 24,303,362 Percentage of placed PE reads:94.5% Num. of uniquely placed SE reads:23,229,651 Percentage of placed SE reads:90.6% Detection results: Number of deletions: 5,816 Number of detected deletions: 5,668(97.5%) Number of false positives:135 (2.3%) Number of insertions: 5,816 Number of detected insertions:5,458(93.8%) Number of false positives:15(0.26%) Indel Detection P.Faciparum 3D7 Simulations

Availability ftp://ftp.sanger.ac.uk/pub/zn1/ssaha_pileup/ More information: ftp://ftp.sanger.ac.uk/pub/zn1/ssaha_pileup/ssaha_pi leup-readme

FuzzyPath and Assemblies from Mixed Solexa/454 Datasets to Extremely GC Biased Genomes

Sequence Reconstruction - Hamiltonian path approach S=(ATGCAGGTCC) S=(ATGCAGGTCC) ATG -> TGC -> GCA -> CAG -> AGG -> GGT -> GTC -> TCC ATG AGG TGC TCC GTC GGT GCA CAG Vertices: k-tuples from the spectrum shown in red (8); Edges: overlapping k-tuples (7); Path: visiting all vertices corresponding to the sequence.

Sequence Reconstruction - Euler path approach Vertices: correspond to (k-I)-tuples (7); Edges: correspond to k-tuples from the spectrum (8); Path: visiting all EDGES corresponding to the sequence. AT GT CG CA GC TG GG ATGCGTGGCA ATGGCGTGCA ATGGCGTGCA ATG -> TGG -> GGC -> GCG -> CGT -> GTG -> TGC -> GCA

Assembly Strategy Selexa reads assembler to extend long reads of 1-2Kb Genome/Chromosome Capillary reads assembler Phrap/Phusion forward-reverse paired reads bp known dist ~500 bp bp

Kmer Extension & Repeat Junctions

Handling of Single Base Variations

ACGTAACTAACAGTT ACGTAACTCACAGTT ACGTAACT ACAGTT Fuzzy Kmers Number of Mismatches between Two Kmers

Means to handle repeats: - Base quality - Base quality - Read pair - Read pair - Fuzzy kmers - Fuzzy kmers - Closely related reference - Closely related reference or Sanger reads or Sanger reads Kmer Extension & Repeat Junctions Pileup of other reads like 454, Sanger etc at a repeat junction Consensus

Pileup of Solexa and 454 Reads

Solexa reads: Number of reads: 3,084,185; Finished genome size: 2,007,491 bp; Read length:39 and 36 bp; Estimated read coverage: ~55X; Number of 454 reads:100,000; Read coverage of 454:10X; Assembly features: - contig stats Total number of contigs: 73; Total bases of contigs: 1,999,817 bp N50 contig size: 62,508; Largest contig:162,190 Averaged contig size: 27,394; Contig coverage over the genome: ~99 %; Contig extension errors: 2 Mis-assembly errors:3 S.Suis P1/7 Solexa/454 Assembly

Solexa reads : Number of reads: 6,000,000; Finished genome size: ~4.8 Mbp; Read length:2x37 bp; Estimated read coverage: ~92.5 X; Insert size: 170/ bp; Assembly features: - contig stats Solexa454 Total number of contigs: 75;390 Total bases of contigs: 4.80 Mbp4.77 Mb N50 contig size: 139,35325,702 Largest contig:395,600 62,040 Averaged contig size: 63,96912,224 Contig coverage on genome: ~99.8 %99.4% Contig extension errors: 0 Mis-assembly errors:04 Salmonella seftenberg Solexa Assembly from Pair-End Reads

Solexa reads : Number of reads: 7,055,348; Finished genome size: 5.35 Mbp; Read length:2x36bp; Estimated read coverage: ~95X; Insert size: 170/ bp; Assembly features: - contig stats Total number of contigs: 168; Total bases of contigs: 5.19 Mbp N50 contig size: 85,886; Largest contig:337,768 Averaged contig size: 30,886; Contig coverage over the genome: ~99 %; Contig extension errors: 1 Mis-assembly errors:2 E.Coli strain 042 Assembly

Solexa reads : Number of reads: 6,346,317; Finished genome size: 4.7 Mbp; Read length:33 bp; Estimated read coverage: ~40 X; Shredded reference of SpA: 10X; Assembly features: - contig stats Total number of contigs: 66; Total bases of contigs: 4,615,704 bp N50 contig size: 168,793; Largest contig:401,700 Averaged contig size: 69,934; Contig coverage over the genome: ~98 %; Contig extension errors: 0 Mis-assembly errors:2 Salmonella delhi5 Solexa Assembly Guided by A Close Reference

The Malaria Genome Project

library organismread lengthMb sequencegenomemean generatedsize (Mb)coverage PCR-free B. pertussis ST242 x PCR-free E. coli 0422 x PCR-free P. falciparum 3D72 x PCR-free B. pertussis ST242 x PCR-free P. falciparum 3D72 x PCR-free E. coli 0422 x standard-245 P. falciparum 3D72 x standard-368 P. falciparum 3D72 x standard-851 P. falciparum 3D72 x standard-883 P. falciparum clin2 x Datasets with Various GC Content GC 68.0% 50.5% 19.0% 50.8% 19.0% 68.0% 19.0%

Solexa reads :2x36 bp2x76 bp Number of reads: 14.0m9.77m Finished genome size: 23 Mbp23 Mbp Estimated read coverage: 43x64x Insert size: 170 bp170 bp Assembly features: Total number of contigs: 26, Total bases of contigs: 19.2 Mbp21.1 Mb N50 contig size: Largest contig: Averaged contig size: Contig coverage on genome: ~83.5 %91.7% Contig extension errors: ?? Mis-assembly errors:?? Malaria 3D7 Assemblies

Acknowledgements:  Jim Mullikin  Tony Cox – Illumina, UK  Tony Cox – Sanger Institute  Adam Spargao,  Yong Gu  Ben Blackburne  Hannes Ponstingl  Daniel Turner  Michael Quail  Jane Rogers  Richard Durbin