Zubair Khalid

Virologist/Molecular Biologist | Veterinarian | Bioinformatician

Conventional & Molecular Virology • Vaccine Development • Computational Biology

Dr. Zubair Khalid is a veterinarian and virologist specializing in conventional and molecular virology, vaccine development, and computational biology. Dedicated to advancing animal health through innovative research and multi-omics approaches.

Dr. Zubair Khalid - Veterinarian, Virologist, and Vaccine Development Researcher specializing in Computational Biology, Multi-omics, Animal Health, and Infectious Disease Research

Category: Guides

Short Read Genome Assembly

Short read genome assembly is the computational reconstruction of a genome sequence from millions of short DNA fragments typically 100 to 300 base pairs long, produced by platforms such as Illumina. This guide provides a rigorous, practical framework for bioinformaticians, genomics researchers, and graduate students who need to assemble genomes from short reads. The process is built on algorithms that piece overlapping reads into longer contiguous sequences, but success depends on careful design choices and quality control. NCBI Bookshelf offers authoritative reference materials, while EMBL EBI Training provides guided tutorials for beginners and advanced users alike.

Whether you are assembling a bacterial genome or a complex eukaryotic genome, short read assembly remains a workhorse technique in genomics. The two dominant algorithmic approaches are overlap layout consensus (rarely used for short reads) and de Bruijn graphs. The de Bruijn graph method breaks reads into k mers and finds paths through the graph to produce contigs. Understanding these fundamentals is essential before you start a project.

At a Glance

Aspect Key Information
Input Short paired end or single end reads (FASTQ files)
Output Contigs and scaffolds (FASTA files)
Main algorithms De Bruijn graph, overlap layout consensus
Typical tools SPAdes, MEGAHIT, ABySS, Velvet, SOAPdenovo2
Computing demands Memory intensive for large genomes, requires cluster or server
Key metrics N50, L50, total assembly size, BUSCO completeness
Common pitfalls Wrong k mer size, inadequate coverage, contamination
Best for Bacterial genomes, resequencing, small eukaryotic genomes

Core Concepts and Decision Criteria

Read Coverage and Sequencing Depth

Coverage is the average number of reads that align to each base of the genome. For short read assembly, 30x to 50x coverage is a common target for bacterial genomes, while eukaryotic genomes may require 50x to 100x. Galaxy Training Network offers example workflows that demonstrate how to calculate coverage from raw read counts. Low coverage leads to gaps, while extremely high coverage can inflate memory usage without improving assembly quality.

K mer Selection

The k mer length is the most critical parameter in de Bruijn assembly. A k mer that is too short creates graph tangles from repetitive regions. A k mer that is too long reduces coverage depth because fewer reads will contain the exact k mer. Many modern assemblers use multiple k mers in a single run. For example, SPAdes iterates over several k mer sizes and merges the best results. Bioconductor packages like kmer help you analyze optimal k mer frequencies before assembly.

Genome Complexity and Ploidy

Simple haploid genomes (bacteria, mitochondria) assemble more easily than diploid or polyploid genomes. Heterozygous regions cause branching in the de Bruijn graph as the assembler tries to represent both alleles. For diploid genomes, some assemblers (Platanus, HapCol) are designed to separate haplotypes. NCBI Sequence Read Archive contains thousands of datasets that illustrate the range of genome complexities you may encounter. A mitochondrial genome assembly, as described for Ligusticum chuanxiong Assembly of the complete mitochondrial genome of Ligusticum chuanxiong and its evolutionary implications, is relatively straightforward because the genome is small and circular.

Decision Matrix for Assembler Selection

Condition Recommended Approach
Small genome (< 100 Mb) SPAdes or MEGAHIT
Large genome (> 1 Gb) ABySS or SOAPdenovo2 with distributed memory
Metagenomic sample MEGAHIT or metaSPAdes
Highly repetitive genome Use longer reads or linked reads as supplement
Diploid heterozygous genome Consider Platanus or dipSPAdes

Practical Workflow and Implementation Steps

Step 1: Raw Read Quality Control

Start by downloading your data from NCBI Sequence Read Archive or your local sequencing facility. Run FastQC to assess per base quality, GC content, adapter contamination, and duplicate levels. Trim low quality bases and remove adapter sequences with tools like Trimmomatic or cutadapt. Bioconductor provides the ShortRead and Rqc packages for quality assessment within R environments. After trimming, re run FastQC to confirm improvement.

Step 2: Error Correction

Short read sequencing introduces substitution and indel errors. Error correction software (e.g., BayesHammer in SPAdes, or standalone tools like Lighter) can reduce the impact of errors on the de Bruijn graph. This step is optional but strongly recommended for genomes below 500 Mb. Overly aggressive correction can remove legitimate rare k mers, so check the corrected reads against a small reference if available.

Step 3: Assembly with De Bruijn Graph

For most projects, SPAdes is a robust choice. Run with default parameters for bacterial genomes, adjusting k mer range based on read length (e.g., 21, 33, 55, 77 for 150 bp reads). For larger genomes, consider MEGAHIT which is memory efficient. Example command:

spades.py -1 reads_1.fastq -2 reads_2.fastq -o assembly_output -k 21,33,55,77 --careful

The careful flag reduces mismatches and short indels. For metagenomic samples, the same assemblers can be used, but a recent example assembled a complete bacterial endosymbiont from citrus leaf metagenome A complete Candidatus walczuchella monophlebidarum genome assembled from citrus leaf metagenomic sequences.

Step 4: Scaffolding

After contig assembly, use paired end information to join contigs into scaffolds. Tools like SSPACE or the built in scaffolder in SPAdes can be used. For genomes with mate pair libraries (larger insert sizes), scaffolding becomes more powerful. The contig graph may also be visualized with Bandage to inspect potential misassemblies.

Step 5: Gap Filling

Scaffolds often contain gaps (Ns) where the assembler could not resolve the sequence. Gap filling tools like GapFiller or Sealer can close some of these gaps. However, repetitive regions may remain unresolved with short reads alone.

Step 6: Polishing

Align reads back to the assembly (using BWA MEM or Bowtie2) and run Pilon or iCORN to correct base errors. This step is critical for generating reference quality genomes, especially for down stream applications like variant calling.

Quality Checks and Validation

Assembly Statistics

Calculate N50 (the contig length at which 50% of the assembly is contained in contigs of that size or larger) and L50 (the number of contigs needed to reach that 50% point). A higher N50 indicates a more contiguous assembly. However, N50 can be inflated by collapsed repeats, so always examine the full histogram of contig lengths. Durum Wheat cv. Svevo Reference Genome Rel.2.0 provides an example where N50 values improved significantly between assembly versions due to better repeat resolution.

Completeness Assessment

Using BUSCO (Benchmarking Universal Single Copy Orthologs) is the gold standard. Run BUSCO against the appropriate lineage dataset (e.g., bacteria, embryophyta, vertebrata). A completion percentage above 90% indicates a good assembly. Avoid using total assembly size as the sole metric, because contaminant sequences can artificially inflate the size.

Read Mapping Consistency

Map reads back to your assembly with BWA MEM. Compute the mapping rate (should be > 95% for a good assembly). Check for regions with zero coverage, which may indicate collapsed repeats or chimeric joins.

Common Mistakes and How to Avoid Them

  1. Using a single k mer. Assemble with multiple k mer sizes and let the assembler pick the best paths. The k mer spectrum from tools like KmerGenie can guide selection.

  2. Ignoring coverage depth outliers. Very high coverage regions often correspond to repetitive elements that the assembler collapses. Inspect coverage plots with IGV and consider masking these regions before downstream analysis.

  3. Insufficient memory allocation. Large genomes require hundreds of GB of RAM. Use a cluster or cloud instance with enough memory. Monitor memory usage during assembly to avoid crashes.

  4. Not removing contaminant sequences. When assembling a target genome from a metagenomic sample, contaminants can form separate contigs that confuse the analysis. Use tools like BlobTools or Kraken to bin contigs taxonomically. The grapevine clonal evolution study A dual genome methylome map of clonal evolution in grapevine used careful filtering to distinguish the host genome from associated microbial genomes.

  5. Over polishing with Pilon. Running Pilon too aggressively can introduce false corrections. Use only one or two iterations and compare the original assembly with the polished one.

Limits and Uncertainty of Interpretation

Short read genome assembly has inherent limitations. Repetitive regions longer than the read length are difficult to resolve, causing collapsed repeats or breaks. Heterozygous regions in diploid genomes produce bubbles in the de Bruijn graph that may be incorrectly merged. Structural variants such as inversions, duplications, and large deletions are often missed or misassembled. For example, the inversion aware transcriptome assembly method Invas Invas: an inversion aware method for transcriptome assembly highlights how standard assemblers can misrepresent such rearrangements. The mucin pangenome study Complex structural variation, phylogeny, and disease associations of the mucin pangenome emphasizes that complex repetitive regions require long read or optical mapping data for accurate resolution.

Do not assume that a short read assembly is complete or error free. Always validate with orthogonal methods: PCR of selected regions, comparison with long read assemblies, or genetic mapping. Uncertainty is highest in centromeres, telomeres, and segmental duplications. For many applications, a draft assembly with high gene completeness is sufficient, but for structural variant analysis or precise genome comparisons, short reads should be supplemented with long reads or Hi C data.

Frequently Asked Questions

What is the minimum coverage needed for a short read genome assembly?

For a bacterial genome, 30x coverage often yields a good draft. For eukaryotic genomes, aim for at least 50x. Very high coverage (> 100x) can be detrimental because it increases memory use and may not improve continuity.

How do I choose the k mer size for my assembler?

Start by running KmerGenie on your trimmed reads to identify the k mer value with optimal peak coverage. Then set the assembler to use a range of k mers around that value (e.g., from 21 to 77 in steps of 12 for 150 bp reads). The automated multik mer approach in SPAdes works well for most projects.

Can short reads assemble repetitive regions?

Short reads can only resolve repeats that are shorter than the read length or that contain unique flanking sequences. Long repeats (e.g., ribosomal DNA arrays, centromeres) typically remain collapsed or fragmented. For complete repeat resolution, you need long reads from PacBio or Oxford Nanopore.

What is scaffolding and why is it important?

Scaffolding orders and connects contigs into longer sequences using paired end or mate pair information. It reduces the number of fragments and reveals the genome structure. Without scaffolding, a genome may appear as thousands of small contigs that are difficult to analyze for gene order or synteny.

References and Further Reading

Related Articles