Genome Assembly: Preliminary Results

Transcription

1 Genome Assembly: Preliminary Results February 3, 2014 Devin Cline Krutika Gaonkar Smitha Janardan Karthikeyan Murugesan Emily Norris Ying Sha Eshaw Vidyaprakash Xingyu Yang

2 Topics 1. Pipeline Review 2. Input data : Samples from CDC 3. Preprocessing 4. Assemblers 5. Preliminary data assembler comparison 6. Visualization 7. The big question: Is it W or C? 8. Revisit pipeline 9. Lab Exercise

3

4 Input Data Number of Single-end reads: 25 Number of Pair-end reads: 25 Number of W serogroup: 23 Number of A serogroup: 1 Number of X serogroup: 1

5

6 Sample Selection Three W serogroups (M22189, M22748, M25543) One A serogroup (M22813) One X serogroup (M24695)

7 Preprocessing Trim start of reads Remove duplicate reads Trim end of reads* Error correction* *Under evaluation

8 Trimming Low Quality Ends of Reads Algorithm: Move window across read until average quality score is below threshold, then remove rest of read from then on. Tools: CG Pipeline run_assembly_trimclean.pl -i <in fastq> -o <out fastq> FastX fastq_quality_filter -Q33 -i <in fastq> -o <out fastq> PRINSEQ prinseq-lite -fastq <in fastq> -out_good <out file name base> -trim_qual_right 30 -trim_qual_type mean -trim_qual_window 10

9 Error Correction Plan has been changed for practical reasons. Tool Algorithm Comments Replacement Coral MSA Works well N/A HiTEC Suffix Tree Requires reads of same length Redeem K-Spectrum Awkward output format Reptile K-Spectrum Awkward output format RACER (newer tool from same developer) N/A BLESS (brand new!)

10 Error Correction Old Plan New Plan Tool Algorithm Tool Algorithm Coral MSA Coral MSA HiTEC Suffix Tree RACER Suffix Tree Redeem K-Spectrum BLESS K-Spectrum Reptile K-Spectrum

11 Error Correction: MSA Algorithm: Align reads and take consensus as an attempt to remove sequencing errors. Tool: Coral coral -fq <in fastq> -o <out fastq>

12 Error Correction: K-Spectrum Algorithm: Replace infrequent k-mers with similar frequent k-mers as an attempt to remove sequencing errors. Tool: BLESS bless -read <in fastq> -prefix <out base file name> -kmerlength 21 (Note that the k-mer length used here differs from the k-mer length used in de novo assembly) Heo Y, Wu XL, Chen D, Ma J, Hwu WM. BLESS: Bloom-filter-based Error Correction Solution for High-throughput Sequencing Reads. Bioinformatics. 2014

13 Error Correction: Suffix Tree Algorithm: Build special data structure (suffix tree) to represent all k-mers present in the reads, replace k-mers found infrequently with neighbors in the tree that are found more frequently. Tool: RACER RACER <in fastq> <out fastq> <estimated genome size> Ilie L, Molnar M. RACER: Rapid and accurate correction of errors in reads. Bioinformatics. 2013;29(19):

14 Assembler Overview Reference: BWA MOSAIK De novo: ABySS MaSuRCA SOAPdenovo2 SPAdes Velvet

15 BWA Reference aligner Commands: # Creating an index/database of reference bwa index -a is <Fasta of Reference> # Aligning reads to reference bwa aln <Fasta for reference> <Fastq of Sample> > <Fastq of sample>.sai # Creating sam files bwa samse <Fasta of Reference> <Fastq of sample>.sai > <Fastq of sample>.sam # Transforming sam file into bam file samtools view -S <Fastq of sample>.sam -b -o <Fastq of sample>.bam # Sorting bam file and creating index of the bam file samtools sort <Fastq of sample>.bam <Fastq of sample> samtools index <Fastq of sample>.bam # Create the consensus samtools mpileup -uf <Fasta of Reference> <Fastq of sample>.bam bcftools view -cg - vcfutils.pl vcf2fq > <Fastq of sample>.cons.fq

16 Mosaik-Aligner Reference aligner Commands: # Creating an index/database of reference MosaikBuild -fr <Reference.fasta> -oa <Reference.dat> # Convert the sample file to mosaik readable file MosaikBuild -q <Sample.fastq> -st <Technology of sequencing> -out <Sample.mkb> #Align the sample file to the reference MosaikAlign -in <Sample.mkb> -out <Sample.mka> -ia < Reference.dat> Mosaik aligner produces two file : Sample.mkb.bam and Sample.mkb.stat #Sorting bam file and creating index of the bam file samtools sort <Sample.mkb.bam> <Samplesort.bam> samtools index <Samplesort.bam> # Create the consensus samtools mpileup -uf <Reference.dat> <Samplesort.bam> bcftools view -cg - vcfutils.pl vcf2fq > <Sample.cons.fq>

17 ABySS de Bruijn assembler Commands: Paired-end reads: abyss-pe k=64 in= pair_file_1.fastq pair_file_2.fastq Single-end reads: abyss-pe k=64 se= single.fastq Both end reads: abyss-pe k=64 lib= pair pair= pair_file_1.fastq pair_file_2.fastq se= single.fastq Results: abyss-fac contig_file.fa

18 MaSuRCA de Bruijn and OLC Commands: Creation of config.txt Includes all parameters including location of paired-end files & kmer length masurca config.txt assemble.sh Output: genome.fasta - scaffold of contigs

19 SPAdes de Brujin Commands: spades.py --only-assembler --careful [options] -o output_dir Options: --pe<#>-1 <forward reads> --pe<#>-2 <reverse reads> --pe<#>-s <unpaired reads> -k int Output is contigs.fa and scaffolds.fa

20 SOAPdenovo2 SOAPdenovo2 usage SOAPdenovo-127mer -s configure_file -K kmer_size An Example of Config file. Insert Length is obtained by map pair end reads to Fam18 with bowtie2

21 Velvet de Bruijn assembler Commands: velveth Single-end reads:./velveth output_dir kmer -fastq -short <SE reads file> Paired-end reads:./velveth output_dir kmer -fastq -shortpaired -separate <PE reads file 1> <PE reads file 2> Single-end & Paired-end reads:./velveth output_dir kmer -fastq -short velvetg <SE reads file> -shortpaired -separate <PE reads file 1> <PE reads file 2>./velvetg output_directory -exp_cov auto -cov_cutoff auto Final output is contigs.fa which are the scaffold files

22 Preliminary Results

23 Assembly Score Determine quality of assembly Calculated using information generated from QUAST

24 Preprocessing

25 Compare Different Trimmers *Single end data only, using SOAP De Novo

26 Different Error Correction Tools *Single end data only, using SOAP De Novo

27 Reference Assembly

28 Mosaik- Reference Aligner single-end reads Sample N50 No. of contigs Total length Assembly score M M M22831(A) M24695(X) No reference No reference No reference No reference M

29 Bwa- Reference Aligner single-end reads Sample N50 No. of contigs Total length Assembly score M M M22831(A) M24695(X) No reference No reference No reference No reference M

30 De novo Assembly

31 Choose the best Kmer Size for each sample Run SOAPdenovo (This assembler runs much faster than others) For all the samples Use kmer size from 51~97

32 Choose the best Kmer Size for each sample The higher, the better. Best Kmer Size: M22189: kmer 85 M22748: kmer 65 M22813: kmer 51 M24695: kmer 59 M25543: kmer 81

33 Paired-End: Sample M22189 Assembler N50 No. of contigs Total length of contigs Assembly score ABySS MaSuRCA SOAPdenovo SPAdes Velvet

34 Comparison of Assemblers Single End

35 Comparison of Assemblers Pair End

36 Comparison of Assemblers Pair End + Single End

37 Comparison of Assemblers

38 Visualization Each of these block outlines surrounds a region of the genome sequence that aligned to part of another genome, and is presumably homologous and internally free from genomic rearrangement. Regions outside blocks lack detectable homology among the input genomes. Inside each block Mauve draws a similarity profile of the genome sequence.

39 Gene transfer between Serogroup C and Serogroup W : Capsule Switching A high mapping percent (>85%) to serogroup C was observed for most of the samples. We investigated the assemblies to find 100% identical Syn G gene(unique to Serogroup W135) but none for SynE gene (unique for serogroup C) in the de-novo assemblies We tried to justify that a certain amount of gene transfer has occurred by the phenomenon of Capsule Switching between the two serogroups.

40 Mapping rate of short reads against reference seq of A, C, W

41 Genetic orientation of capsule biosynthesis and transport loci in Neisseria meningitidis

42 Blast syng against SOAPdenovo2 assemblies of Serogroup W

43 Visualization:Contig Reordering

44

45 Lab Exercise Can be found on the CompGenomics Exercises Wiki page Genome Assembly Lab Exercise Due next week - February 10th