Team I Genome Assembly Group

Introduction and Objectives

In bioinformatics, sequence assembly of a genome is the first of many steps involved to identify and characterize an organism. It may be considered the most important step in the stages of analysis and interpretation (see below) because of the challenge that persist concerning high quality genome assembly and annotation [4].

The basic principle of genome assembly is to note that the more similarity that exists between the end of one read and the beginning of another, the more likely they are to have originated from overlapping stretches of the genome. The output of an assembly is typically a set of ‘‘contigs,’’ which are contiguous sequence fragments, ordered and oriented into ‘‘scaffold’’ sequences, with gaps between contigs within scaffolds representing regions of uncertainty. There are numerous subclasses of assembly problems that can be distinguished by, among other things, the nature of: (1) the reads, (2) the types of sequences being assembled, and (3) The availability of homologous (related) and previously assembled sequences, such as a reference genome or the genome of a closely related species.

Using the most relevant and high-quality tools is important for performing comparative genomics. Importantly, results may have implications in public health that may affect many lives. For the purposes of our project, we are tasked with identifying the source/origin of a given food-borne illness from 50 isolates in FASTQ format that were sequenced using Illumina technology. We are also tasked with developing a workflow for our analysis and using publically available bioinformatic tools that we deem appropriate for identifying our given organism.

Stages of analysis and interpretation of data

1. genome assembly
2. gene prediction
3. functional annotation
4. comparative genomics
5. production of a predictive webserver

Overview of Genome Assembly

Figure 1. General Genome Assembly Workflow (Image retrieved from 101genome)

---

Team Goals

1. To perform quality control on reads before and after assembling the genome:

Before:

• Fastp

After:

• QUAST

2. To evaluate the performance of assembly tools:

• Abyss
• Unicycler
• SPADES
• SKESA
• MaSuRCA
• Velvet

3. To use the best tool to perform de novo assembly based on the 50 isolates.

4. To send off the highest quality result to the gene prediction team.

Team 1 | Genome Prediction

Methods

Proposed Genome Assembly Pipeline

Figure 2. A proposed pipeline for genome assembly based upon class lectures and review of literature.

Raw Sequence Data

Our team was assigned 50 isolates sequenced using an Illumina sequencer. All isolates are in FASTQ format. FASTQ is a standard format that contains the read sequence and a quality for every base.

Quality Control & Trimming

It is important to check quality of sequences prior to proceeding to assembly including checking the quality of the average base quality score per read, the GC content distribution and identification of the most duplicated reads. Proceeding with assembly of the sequences without checking for the quality of the reads will lead the misinterpretation of results. For the purposes of our project we will be using fastp() for quality control analysis as well as read trimming. We will use Fastp because it includes most features of similar tools used for quality control and trimming (including FASTQC + Cutadapt + Trimmomatic + AfterQC), all while running 2 to 5 times faster than any of them alone.

• Our threshold Minimum quality score: 22

• Our Sliding window: 10

• We will trim reads from both ends

Figure 3. Fastp output: Optimal sliding window and minimum quality score parameters for our analysis

Assembly

An important decision when approaching the assembly stage of genome is to determine whether performing assembly via a reference genome or de novo assembly. Reference assembly refers to aligning the sequences to a known genome. De novo assembly refers to aligning the sequences via overlapping groups called contigs to build a novel genome. There are pros and cons to each approaches. For assembly to a reference genome, the assembly can be done relatively quickly. It is also good for single nucleotide variants (SNV) and small indels. However the reference genome approach is limited length for feature detection and requires a reference that is close to the sequenced data. The de novo assembly approach is good for completely new sequences not present in the reference. However it can be slow and has high infrastructure requirements and is bad at hiding raw data limitations.

It is important to understand characteristics of the biological data you are working with to make certain decisions when deciding to use a bioinformatic tool. The genome used for our project is not human and likely prokaryotic (due to basepair count). Unlike the human genome, that is about 99.9% the same across species, bacterial and viral genomes can change relatively quickly, making the approach of genomic assembly via a reference genome not ideal due to the potential loss of information when mapping occurs. For this reasons, we will proceed in assembling the sequence reads using a de novo assembly approach.

Figure 4. Reference genome assembly diagram.

de Novo Assembly

Figure 5. De nova genome assembly diagram.

De novo assembly is the most common type of genome assembly for short read sequences. It involves reconstructing entire genome from overlapping sequence reads. The quality depends on the size of the reads and number of gaps between them. Most tools use either de Bruijn graphs (Eulerian) or Overlap graphs (Hamiltonian) as their algorithms. De novo assembly can generate new and accurate reference sequences, even for complex genomes.

Qualitative evaluation of genome assembly tools

Rationale for (Assembler) tool selection

Based on literature that provide an unbiased assessment of pro/cons of various genome assembly tools. (NOTE: Not all pros/cons are displayed)

• • GAGE (Genome Assembly Gold-standard Evaluations)
    • A study designed to provide a snapshot of how the latest genome assemblers compare on a sample of large-scale next-generation sequencing projects.
    • Helps to answer questions like: Which assembly software will produce the best results?

• • Assemblathon I
    • A (critical) competitive assessment of de novo short read assembly methods.
    • Aims to comprehensively assess the state of the art in de novo assembly methods when applied to current sequencing technologies.

Tool	Pros	Cons
ABySS	Uses a distributed k-mer hash table, making it more RAM-efficient	Produces low N50 contig and scaffold numbers
ALLPATHS-LG	High contig N50 value	Speed is slow
SPADES	Large contigs	Generates small contigs if coverage is low
SKESA	High sequence quality and contiguity	Does not have a built-in scaffolding tool
Velvet	Great for sequences rich in repeat segments	Small N50 contig size
Unicycler	Assembles larger contigs with fewer misassemblies	Runtime is long
MaSuRCA	Usable for large genomes, High NG50 contig size	Speed is slow

Table 1. Qualitative analysis of select genome assembly tools.

Assembly Validation Criteria

It is generally the case that the right answer to an assembly problem is unknown. Understandably therefore, common methods for assessing assembly quality are needed. Below is our evaluation criteria for determining the best genome assembly tool to use for our project.

Metric	Description
N50	The minimum contig length crossing the 50% threshold of the total assembled size of the genome.
L50	The smallest number of contigs whose summed length is N50.
Number of Contigs	The total number of contigs in the assembly
Largest Contigs	The length of the largest contig in the assembly
Total Length	The total number of bases in the assembly

Table 2. Evaluation criteria of genome assembly tools.

QUAST

Quast is a quality assessment tool used for genome assembly. QUAST can evaluate assembly quality with or without a reference genome, so that it is useful for new species which do not yet have a finished reference genome [1].

Results

Quantitative Comparison of Genome Assemblers

Using our evaluation criteria, 5 randomly selected isolates (CGT1294, CGT1893, CGT1966, CGT1977, and CGT1990) representing 10% of our 50 total isolates were used to determine which genome assembly tool performed best. Analysis of contigs.fa files for each tool was assessed using 'QUAST'. The best tool for downstream analysis was chosen and was used for further analysis on the remaining 45 isolates.

Number of Contigs

Figure 6.

---

Largest Contigs

Figure 7.

---

Total Length

Figure 8.

---

N50

Figure 9.

---

L50

Figure 10.

---

Assemblers Eliminated from Consideration

• Velvet

• ABySS

• SPADES

• SKESA

Final Assemblers

• MaSuRCA

MaSuRCA is whole genome assembly software. It combines the efficiency of the de Bruijn graph and Overlap-Layout-Consensus (OLC) approaches. MaSuRCA can assemble data sets containing only short reads from Illumina sequencing or a mixture of short reads and long reads.

• Unicycler

Unicycler is designed specifically for hybrid assembly which uses both short-read and long-read sequencing data of small (e.g., bacterial, viral, organellar) genomes. Unicycler builds an initial assembly graph from short reads using the de novo assembler SPAdes and then simplifies the graph using information from short and long reads.

MaSuRCa and Unicycler comparison

Identification of Organism

Figure 10. Image of E. coli" | Image retrieved from: (Cleveland Clinic)

Identity of organism: E. coli

• • Genome size: about 4.7 Mbp
  • Average GC content: About 50%
  • Disease state: Symptoms vary from person to person. Often involves severe stomach cramps, diarrhea (often bloody), and vomiting. Some people may have a fever, which usually is not very high (less than 101˚F/38.5˚C). Most people get better within 5 to 7 day. Some infections are very mild, but others are severe or even life-threatening [4].

Final Genome Assembly Pipeline

Conclusion

In conclusion, we determined Unicycler and MaSuRCA were the optimal tools to be used for assembly. Through blast anaylsis we determined the species of the pathogen to be Escherichia coli. We will pass on our contigs to the gene prediction group for analysis.

In-Class Presentations

• Genome Assembly Background and Strategy: File:Team 1 Genome Assembly Presentation 1.pdf

• Genome Assembly Final Results:

References

1. Alexey Gurevich, Vladislav Saveliev, Nikolay Vyahhi, Glenn Tesler, QUAST: quality assessment tool for genome assemblies, Bioinformatics, Volume 29, Issue 8, 15 April 2013, Pages 1072–1075, https://doi.org/10.1093/bioinformatics/btt086

2. Bankevich A, Nurk S, Antipov D, et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J Comput Biol. 2012;19(5):455–477. doi:10.1089/cmb.2012.0021

3. Butler, Jonathan et al. “ALLPATHS: de novo assembly of whole-genome shotgun microreads.” Genome research vol. 18,5 (2008): 810-20. doi:10.1101/gr.7337908

4. CDC, (2017) Centers for Disease Control and Prevention, National Center for Emerging and Zoonotic Infectious Diseases (NCEZID), Division of Foodborne, Waterborne, and Environmental Diseases (DFWED)

5. Earl, Dent et al. “Assemblathon 1: a competitive assessment of de novo short read assembly methods.” Genome research vol. 21,12 (2011): 2224-41. doi:10.1101/gr.126599.111

6. Maccallum, Iain et al. “ALLPATHS 2: small genomes assembled accurately and with high continuity from short paired reads.” Genome biology vol. 10,10 (2009): R103. doi:10.1186/gb-2009-10-10-r103

7. Miller, Jason R et al. “Assembly algorithms for next-generation sequencing data.” Genomics vol. 95,6 (2010): 315-27. doi:10.1016/j.ygeno.2010.03.001

8. Pritt, J., Chen, N. & Langmead, B. FORGe: prioritizing variants for graph genomes. Genome Biol 19, 220 (2018). https://doi.org/10.1186/s13059-018-1595-x

9. Quainoo, S., Coolen, J.P., Hijum, S.A., Huynen, M.A., Melchers, W.J., Schaik, W.V., & Wertheim, H.F. (2017). Whole-Genome Sequencing of Bacterial Pathogens: the Future of Nosocomial Outbreak Analysis. Clinical microbiology reviews, 30 4, 1015-1063 .

10. Rahman, A., Pachter, L. CGAL: computing genome assembly likelihoods. Genome Biol 14, R8 (2013). https://doi.org/10.1186/gb-2013-14-1-r8

11. Salzberg, Steven L et al. “GAGE: A critical evaluation of genome assemblies and assembly algorithms.” Genome research vol. 22,3 (2012): 557-67. doi:10.1101/gr.131383.111

12. Shifu Chen, Yanqing Zhou, Yaru Chen, Jia Gu; fastp: an ultra-fast all-in-one FASTQ preprocessor, Bioinformatics, Volume 34, Issue 17, 1 September 2018, Pages i884–i890, https://doi.org/10.1093/bioinformatics/bty560

13. Sohn, Jang-il; Nam, Jin-Wu. “The present and future of de novo whole-genome assembly”, Briefings in Bioinformatics, Vol 19.1 (2018). doi.org/10.1093/bib/bbw096

14. Souvorov A., Agarwala R., & Lipman D.J. SKESA: strategic k-mer extension for scrupulous assemblies. Genome Biology. 2018; 19(1). doi:10.1186/s13059-018-1540-z

15. Tanja Magoc, Stephan Pabinger, Stefan Canzar, Xinyue Liu, Qi Su, Daniela Puiu, Luke J. Tallon, Steven L. Salzberg, GAGE-B: an evaluation of genome assemblers for bacterial organisms, Bioinformatics, Volume 29, Issue 14, 15 July 2013, Pages 1718–1725, https://doi.org/10.1093/bioinformatics/btt273

16. Zerbino, D., & Birney, E. (n.d.). Velvet: de novo assembly using very short reads. Hinxton: European Bioinformatics Institute.