Genome properties Henrik Lantz NBISSci Life LabUppsala University

Genome properties Henrik Lantz - NBIS/Sci. Life. Lab/Uppsala University

Organisms are different, and so are assembly projects

Genome properties • • • Genome size Heterozygosity levels Repeat-content GC-content Secondary structure Ploidy level

Genome size • Genome sizes range from 100 kbp to 150 Gbp • The larger the genome, the more data is needed to assemble it (>50 x usually) • Compute needs grow with increased amount of data (running time and memory) • Note that larger genomes do not necessarily have to be harder to assemble, although empirically this is often the case

Heterozygosity (Slide by Torsten Seeman, Victorian Life Sciences Computation Initiative)

Highly heterozygous fungus (Zheng et al. (2013) Nature Com. )

Heterozygosity • Highly heterozygous regions tend to be assembled separately • Homologous regions existing in multiple copies in the assembly • Downstream problems in determining orthology for gene based analyses, comparative genomics etc.

Effect of heterozygosity on assembly size (Pryszcz and Gabaldon (2016) Nucl. Acids. res. )

Repeats • Identical, or near identical, regions occurring in multiple copies in a genome (Istvan et al. (2011), PLo. S ONE)

Repeats • Low complexity regions Regions where some nucleotides are overrepresented, such as in homopolymers, e. g. , AAAAA, or slightly more complex, e. g. , AAATAAAAAGAAAA • Tandem repeats A pattern of one or more nucleotides repeated directly adjacent to each other, e. g. , AGAGAGAGAG 2 -5 nucleotides - microsatellites (e. g. , GATAGATA) 10 -60 nucleotides - minisatellite • Complex repeats (transposons, retroviruses, segmental duplications, r. DNA, etc. )

How repeats can cause assembly errors Mathematically best result: C R A B

Repeat errors Overlapping non-identical reads Wrong contig order Collapsed repeats and chimeras Inversions

When can I expect repeats to cause a problem? • Always… • Much more common in eukaryotes, in particular plants and many animals • Several conifers have a repeat content of ~75%, mostly simple repeats -> huge genomes

How to deal with repeats • Long range information, e. g. , long reads or paired reads with long insert sizes R 1 Short reads R 2

How to deal with repeats • Long range information, e. g. , long reads or paired reads with long insert sizes Long reads

Genome properties • GC-content – Regions of low or high GC-content have a lower coverage (Illumina, not Pac. Bio) • Secondary structure – Regions that are tightly bound get less coverage • Ploidy level – On higher ploidy levels you potentially have more alleles present

Additional complexity • Size of organism – Hard to extract enough DNA from small organisms • Pooled individuals – Increases the variability of the DNA (more alleles) • Inhibiting compounds – Lower coverage and shorter fragments • Presence of additional genomes/contamination – Lower coverage of what you actually are interested in, potentially chimeric assemblies

Project planning Henrik Lantz - NBIS/Sci. Life. Lab/Uppsala University

Sequencing technology comparison Sequencing system Read length Yield Illumina Hi-Seq 2500 2 x 125 bp 180 M read pairs/lane, 28 Gbp/lane Illumina Hi. Seq. X 2 x 150 bp 350 M read pairs/lane, 78 Gbp/lane Illumina Nova. Seq 6000 2 x 150 bp 2000 M read pairs/lane Illumina Mi. Seq Up to 2 x 500 bp 18 M read pairs/lane, 7. 4 Gbp/run Pac. Bio RSII 250 -70000 bp 1 Gb/SMRTcell Pac. Bio Sequel 250 -300000 bp ~10 Gb/SMRTcell Oxford Nanopore 500 -1000000 bp 148 Gb/Flow cell

Error rates and types Sequencing system Error type Error rate Illumina Substitutions 0. 1% Pac. Bio Insertions 0. 001 -12% depending on read length Oxford Nanopore Substitutions, indels 15%

De novo genome project workflow • • • Plan your project! Extract DNA (and RNA) Choose best sequence technology for the project Sequencing Quality assessment and other pre-assembly investigations Assembly validation Assembly comparisons Repeat masking? Annotation

De novo genome project workflow • Plan your project!

What do you want to achieve? Quality Draft genome, split in longer repeats, complete genes, Draft genome, highly fragmented, almost full gene space split genes, partial gene space Effort and cost Fully assembled and phased genome, full gene space

Pilot project? • One lane of Illlumina data is cheap and can be used to investigate genome size, presence of contaminants, and more. • Long read technologies are sensitive to DNA quality issues. Trying several extractions before deciding which one to use can be a good idea. An extraction that gives good QC values (fragment sizes, absorption rates, etc. ), can still fail in sequencing!

Estimate computing resources (Dominguez del Angel et al. , 2018)

Estimate computing resources • What tools do you want to run? • Assembly can be memory-intense. Polishing can also require a lot of memory. A normal Rackham node might not be enough. • Do you have the necessary storage space? • Can you run your tools on several nodes over MPI?

DNA extraction Henrik Lantz - NBIS/Sci. Life. Lab/Uppsala University

Example

Causes of DNA degradation • • • Mechanical damage during tissue homogenization. Wrong p. H and ionic strength of extraction buffer. Incomplete removal / contamination with nucleases. Phenol: too old, or inappropriately buffered (p. H 7. 8 – 8. 0); incomplete removal. Wrong p. H of DNA solvent (acidic water). Recommended: 1: 10 TE for short-term storage, or 1 x. TE for long-term storage. Vigorous pipetting (wide-bore pipet tips). Vortexing of DNA in high concentrations. Too many freeze-thaw cycles (we tested 5, still Ok). Debatable: sequence-dependent

What are the main contaminants? Polysaccharides Lypopolysaccharides Growth media residuals Chitin Protein Secondary metabolites Pigments Growth media residuals Chitin Fats Proteins Pigments Polyphenols Polysaccharides Secondary metabolites Pigments By Olga Vinnere Pettersson, Uppsala Genome Center, Sci. Life. Lab

What do absorption ratios tell us? Pure DNA 260/280: 1. 8 – 2. 0 < 1. 8: Too little DNA compared to other components of the solution; presence of organic contaminants: proteins and phenol; glycogen - absorb at 280 nm. > 2. 0: High share of RNA. Pure DNA 260/230: 2. 0 – 2. 2 <2. 0: Salt contamination, humic acids, peptides, aromatic compounds, polyphenols, urea, guanidine, thiocyanates (latter three are common kit components) – absorb at 230 nm. >2. 2: High share of RNA, very high share of phenol, high turbidity, dirty instrument, wrong blank. By Olga Vinnere Pettersson, Uppsala Genome Center, Sci. Life. Lab Photometrically active contaminants: phenol, polyphenols, EDTA, thiocyanate, protein, RNA, nucleotides (fragments below 5 bp)

DNA quality requirements Some DNA left in the well Sharp band of 20+kb No sign of proteins No smear of degraded DNA No sign of RNA Nano. Drop: Qubit or Picogreen: 260/280 = 1. 8 – 2. 0 260/230 = 2. 0 – 2. 2 10 kb insert libraries: 3 -5 ug 20 kb insert libraries: 10 -20 ug

De novo genome project workflow • Plan your project! • Extract DNA (and RNA) Extract much more DNA than you think you need Also remember to extract RNA for the annotation Single individual and haploid tissue if possible In particular for Illumina mate-pairs data and Pac. Bio, a lot of high molecular weight DNA is critical! • Extracting DNA for de novo assembly is very different from extractions intended for PCR • Do several extractions if possible, and run them on a gel to get an idea of how fragmented the DNA is • Try to remove contaminants from the extractions • •

Effect of insert size on scaffold length (Treangen and Salzberg, 2013)