2015.04.08-Next-generation-sequencing-issues
- 1. Learn from Prac,ce -‐What Tells You about a Problema,c NGS Dongyan Postdoctoral Research Associate Buell Lab/Jiang Lab 2015.4.8
- 2. Sources affec,ng NGS 1. Systema,c varia,on in quality scores across the sequence read 2. Quality trimming and cleaning of raw reads 3. Biases in sequence genera,on driven by base composi,on 4. Contamina,on from known and unknown species other than the sequencing target 5. NGS libraries on assembly quality 6. others 7. ………………………………….
- 4. Per base quality score Forward reads Reverse reads Sample A Sample B
- 5. 200 bp 300 bp 400 bp 500 bp 700 bp 800 bp Library QC using Bioanalyzer Sample A Sample B Adapted from the report generated by Emily Crisovan (Buell lab)
- 6. Cause for the poor base quality for Sample B Illumina flowcells may not handle longer fragments well Bronner et al., 2009
- 7. diol diol 1st cycle denaturation 1st cycle annealing diol diol n=35 total 1st cycle extension diol diol diol diol 2nd cycle denaturation 2nd cycle annealing dioldiol diol Cluster Genera7on: Amplifica7on diol dioldiol 2nd cycle extension Adapted from Robin’s slides
- 8. Per base sequence quality Before cleaning Aaer cleaning Good base quality is the start.
- 9. QUALITY TRIMMING AND ADAPTER REMOVING OF RAW READS
- 10. k-‐mer content (residual adapter sequences) –paired-‐end reads Before cleaning Aaer cleaning • Only happened to paired-‐end libraries with small insert size (<400 bp). • Not happen to paired-‐end libraries with insert size greater than 400 bp.
- 11. k-‐mer content • This is due to the ‘reading through’ a short fragment into the adapter sequence on the other end. • The default threshold of the clip is too high? • ILLUMINACLIP:TruSeq3-‐PE.fa:2:30:10
- 12. k-‐mer content (residual adapter sequences) –mate pair reads Aaer cleaning and grouping reads to categories using NextClip • Those k-‐mers are from the junc,on adapter
- 13. k-‐mer content • Didn’t want to lower down the threshold in case it may clip more than necessary • Used cutadapt and its default selng to remove the residual adapter sequences aaer trimmoma,c and/or NextClip cleaning
- 14. Residual adapter on assembly w/ residual adapter w/o residual adapter Never rush to assembly before you are sure you have a high-‐quality and ‘clean’ read sets!
- 15. BIASES IN SEQUENCE GENERATION DRIVEN BY BASE COMPOSITION
- 16. Biases in sequence genera,on Paired end reads GC%: 33% Mate pair reads GC%: 40%
- 17. CONTAMINATIONS
- 18. Per sequence GC content SGA preQC Sample A Sample B Contamina,ons?
- 19. QC • Map reads back to the assembly • Taxon-‐Annotated Gene Content • MAKER annota,on of the assembly • OrthoMCL analysis
- 20. Mapping reads to the assemblies • Assembled reads using ABySS • Map reads back to the assembly using Bow,e/ 1.0.0 in single end mode allowing 1 mismatch Sample B assembly reads mapped unmapped Sample A 73.37% 26.63% Sample B 60.94% 39.06% Contamina,ons?
- 21. TAGC Sample A Sample B hpps://github.com/blaxterlab/blobology hpps://github.com/mojones/blobsplorer
- 22. TAGC-‐highlighted a phylum Sample A Streptophyta
- 23. TAGC-‐highlighted a phylum Sample B Proteobacteria Streptophyta
- 24. Maker annota,on Data used #con,g>1000bp Sample A assembly 75,417 Sample B assembly 92833 • EST evidence • caa_assembly.fasta (Elsa) • Protein homology evidence: • uniprot_sprot_plants.fasta • TAIR10_pep_20110103_representa,ve_gene_model • Repeat masking-‐default Sample A Sample B Num_of_transcripts 31,234 45,791 Max_len_trans 14,796 29,577 Min_len_trans 28 33 N50 17,253,963 27,945,180 N50 transcript size 1,409 1,498 Average transcript size 1,105 1,221 With help from Kevin Childs
- 25. OrthoMCL analysis • OrthoMCL DB (web-‐based) – hpp://www.orthomcl.org/orthomcl/ – search against predefined sets of orthologous groups from a set of organisms
- 26. OrthoMCL analysis Steps: 1. All-‐vs-‐all BLASTP of the proteins 2. Compute percent match length -‐ Select whichever is shorter, the query or subject sequence. Call that sequence S. -‐ Count all amino acids in S that par,cipate in any HSP. -‐ Divide that count by the length of S and mul,ply by 100. 3. Apply thresholds to blast result. Keep matches with E-‐value < 1e-‐5, percent match length >= 50%. 4. Find poten,al inparalog, ortholog and co-‐ortholog pairs using the Orthomcl Pairs program (These are the pairs that are counted to form the Average % Connec,vity sta,s,c per group). 5. User the MCL program to cluster the pairs into groups. orthomclResults/ 1. orthologGroups a map between your proteins and OrthoMCL groups. 2. paralogPairs reciprocal best hits among those proteins in your genome 3. that were not mapped to OrthoMCL groups 4. paralogGroups the proteins in paralogPairs clustered into groups by the mcl program
- 27. OrthoMCL analysis orthologGroups your_protein, orthomcl_group, seq_id_of_best_hit, evalue_man7ssa, evalue_exponent, percent_iden7ty, percent_match • Downloaded the “category” , “species name”, and “abbrevia,on” info from the website • Used perl scripts to add the corresponding species name and category to the orthologousGroups file • Calculated # of orthologous groups in each category
- 28. OrthoMCL analysis category abbrevia,on Archaea ARCH Bacteria FIRM Bacteria OBAC Bacteria PROT Fungi FUNG Metazoa META other Eukaryota OEUK Pro,st ALVE Pro,st AMOE Pro,st EUGL Viridiplantae VIRI
- 29. Orthologous groups category abbrevia ,on Archaea ARCH Bacteria FIRM Bacteria OBAC Bacteria PROT Fungi FUNG Metazoa META other Eukaryota OEUK Pro,st ALVE Pro,st AMOE Pro,st EUGL Viridiplantae VIRI FIRM: Firmicutes OBAC: Other Bacteria PROT: Proteobacteria Bacteria Pro,st Sample A Sample B Sample A+B
- 30. SEQUENCING LIBRARIES ON GENOME ASSEMBLY
- 31. Assembly Using ABySS • MP libraries improved the assembly Libraries k-mer total# of contigs #contigs>= 500bp #contigs> N50 N50 max sum #N Paired-end reads only New PE (4 libraries) 75 500,224 75,165 9,009 11,748 106,374 367,800,000 281,121 New PE (6 libraries) 75 504,588 74,671 8,886 11,911 106,374 367,800,000 297,396 Paired-end and mate pair reads New PE (4 libraries) + MP (2 libraries) 75 168,163 31,320 3,088 37,026 289,426 401,000,000 281,121 New PE (6 libraries) + MP (2 libraries) 75 171,733 29,974 3,000 38,350 274,863 401,200,000 297,396
- 32. OTHER THINGS
- 33. SRA • Reads from DRR004446.sra and DRR004447.sra are exactly the same • # Run # of Spots # of Bases Size • 1. DRR004446 14,841,025 2.7G 1.5Gb • # Run # of Spots # of Bases Size • 1. DRR004447 14,841,025 2.7G 1.5Gb
- 34. Take home message • You can’t be over cau,ous with NGS data! • Always do QC before further analysis! hpp://en.wikipedia.org/wiki/DNA_sequencing