Genome Assembly Workflow: How to Plan Short-Read and Long-Read Projects

A genome assembly workflow converts sequencing reads into a reconstructed genome. The choice between short-read and long-read technologies depends on your genome size, complexity, budget, and required contiguity. This guide is for researchers who need to plan a de novo genome assembly project, whether for a bacterium, a plant, or an animal. It compares inputs, strategies, polishing, validation, and data release planning.

At a Glance

Decision Criteria

Genome Size and Complexity

Small bacterial genomes (up to 10 Mb) assemble well with short reads alone. Large genomes (>100 Mb) with repetitive content benefit from long reads. For example, plant genomes with long repeats nearly require long-read contiguity. The Galaxy Training Network provides hands-on tutorials for both approaches (see Galaxy Training Network).

Budget and Timeline

Short-read projects cost less per base, especially if you already have access to an Illumina instrument. Long-read sequencing library preparation is more expensive, and the instruments have higher upfront costs. However, long reads can reduce the need for expensive finishing and gap closure.

Required Assembly Quality

If you need a reference-grade genome with N50 > chromosome length, long reads are essential. For comparative genomics or gene annotation a short-read draft may suffice. EMBL EBI training resources cover how to define quality metrics (see EMBL EBI Training).

Practical Workflow or Implementation Sequence

A complete genome assembly project follows these steps. Adapt the details to your read type.

Step 1: DNA Extraction and Library Preparation

High molecular weight DNA is critical for long-read libraries. Use a gentle extraction method (e.g., agarose plug or column with wide bore pipettes). For short reads, standard column based extraction works. Check for contaminants with spectrophotometry and fluorometry.

Step 2: Sequencing and Raw Data QC

Sequence your library on the chosen platform. For short reads, aim for paired end reads of 150 bp. For long reads, check the Q score distribution and trim adapters. Use tools like FastQC and NanoPack. Store raw data in the NCBI Sequence Read Archive (SRA) after the project (see NCBI Sequence Read Archive).

Step 3: Read Filtering and Error Correction

Short reads: trim low quality bases and remove adapters (Trimmomatic, Cutadapt). Long reads: use Canu's built in correction or run Racon for iterative error correction. Do not skip this step, raw long reads have up to 15% error.

Step 4: Assembly

For short reads: SPAdes is a reliable all in one assembler for bacteria and small eukaryotes. For larger genomes, consider ABySS or Megahit. For long reads: Flye works well for bacteria and small eukaryotes. For large genomes, Canu or Miniasm plus Racon. Hybrid assembly pipelines like Unicycler use short reads to polish long read assemblies.

Step 5: Polishing

Polishing fixes base level errors. For short read assemblies, polish with Pilon or iCORN. For long read assemblies, use Medaka (Oxford Nanopore) or Arrow (PacBio) and then further polish with Pilon using high quality short reads. Polishing is essential because raw long read assemblies have many insertions and deletions. The Bioconductor project offers packages like polish and gmapR for validation (see Bioconductor).

Step 6: Validation

Validate your assembly using multiple independent metrics. Compute contiguity (N50, L50), completeness (BUSCO), and check for misassemblies (QUAST, REAPR), and map reads back to assess coverage uniformity. A published article emphasizes validation centered approaches for bacterial genomes (see PubMed article 42403179). For metagenome assembled genomes, use checkM.

Step 7: Data Release Planning

Prepare your assembly for public release. Submit the final assembly to NCBI GenBank, INSDC members, or EBI. Deposit raw reads in the NCBI Sequence Read Archive (SRA) (see NCBI Bookshelf for submission guidelines). Write a Data Availability Statement for your manuscript. Include assembly accession numbers.

Common Mistakes

Insufficient Coverage

Short read assemblies need at least 50x coverage for bacteria, more for eukaryotic genomes. Long read assemblies require 30x for correction. Under coverage leads to fragmented assemblies and missing genes. Calculate coverage from total base output divided by genome size.

Ignoring GC Bias

Short read libraries can underrepresent GC rich or AT rich regions. Check the GC content distribution in your reads. If biased, consider using PCR free libraries or adding long reads. The validation paper cited above (PubMed 42403179) discusses how GC bias affects accuracy.

Skipping Polishing

Long read assemblies without polishing have error rates of 1,5%. That is unacceptable for most downstream analyses. Always polish, even with the same read set if you have high quality short reads. Use Pilon with Illumina data after a Flye assembly.

Weak Contamination Check

Contamination from host, reagents, or other samples can produce chimeric contigs. Run a tool like Mash or Kraken2 on raw reads and on assembled contigs. The Galaxy Training Network has workflows for decontamination (see Galaxy Training Network). For clinical isolates, pay attention to cross contamination as shown in a recent study on Acinetobacter baumannii (see PubMed 42383857).

Limits and Uncertainty

No assembly is perfect. Repeats longer than the read length remain collapsed or misjoined. High heterozygosity in diploid or polyploid organisms leads to fragmented assemblies without specialized tools (e.g., trio binning or phased assembly). Genome size estimation can be inaccurate if you rely only on kmer counts without a trusted reference. The uncertainty in assembly quality increases with genome complexity. Use multiple validation metrics and be transparent about remaining gaps and errors in your publication.

Frequently Asked Questions

1. What coverage should I aim for with short reads?
For bacterial genomes, 50,75x paired end 150 bp reads usually produces a good draft. For larger eukaryotic genomes, aim for 80,100x. Higher coverage can help resolve repeats but may also introduce errors if the data are biased.

2. Can I combine short and long reads in one assembly?
Yes, hybrid assembly combines the contiguity of long reads with the accuracy of short reads. Tools like Unicycler, Masurca, and SPAdes hybrid mode are popular. They polish long read assemblies with short reads and can close gaps that remain in long read only assemblies.

3. How do I check if my assembly is complete?
Run BUSCO against a lineage specific dataset. This tool counts conserved single copy orthologs. A complete genome should recover >95% of BUSCOs as single copy. Also check that the assembly size is close to the estimated genome size (from flow cytometry or kmer analysis).

4. What should I do if my assembly contains many small contigs?
Small contigs may be plasmids, repeats, or contamination. Blast them against NCBI nt or GenBank. For bacteria, check for circularity of predicted chromosomes and plasmids. Use a gap closing step with tools like GapFiller or gap5.

References and Further Reading

• NCBI Bookshelf: Submitting assemblies to GenBank , authoritative guide on data submission and metadata.
• EMBL EBI Training: Assembly quality assessment , module on metrics and tools.
• Galaxy Training Network: Assembly tutorials , hands on exercises for short read and long read assembly.
• Bioconductor: Genomic interval and assembly packages , software for polishing and validation.
• NCBI Sequence Read Archive: Raw data repository , how to submit your FASTQ files.
• From contiguity to accuracy: Validation centered perspectives on bacterial genome assembly. J Microbiol. 2024. PubMed 42403179 , discusses validation strategies.
• Comprehensive identification of sequence types belonging to Acinetobacter baumannii clonal complexes. Microb Genom. 2024. PubMed 42383857 , example of contamination aware assembly.
• nf core/magmap: Map metatranscriptomes to large collections of genomes. Bioinformatics. 2025. PubMed 42429454 , tool for mapping reads to assemblies.
• Coupled perturbations of gene circuit dynamics by resource competition. Methods Mol Biol. 2025. PubMed 42420739 , example of high quality reference use.
• From candidate genes to omics: Unbiased approaches reshaping arthropod Evo Devo. Genet Mol Biol. 2024. PubMed 42397051 , demonstrates assembly in non model organisms.

Related Articles

• RNA Sequencing Analysis: From FASTQ Files to Biological Questions
• RNA-seq Quality Control: What to Check Before Differential Expression
• How to Plan a Bulk RNA-seq Differential Expression Study
• Single-Cell RNA-seq Workflow: A Practical Analysis Roadmap
• Single-Cell RNA-seq Quality Control: Cells, Genes, and Mitochondrial Reads