How Much Sequencing Coverage Do You Need for a Bacterial Genome?

What coverage means

Coverage is how many times, on average, each base of the genome has been sequenced. It is calculated by dividing the total number of sequenced bases by the genome size. If a 5 Mb genome is sequenced with 250 Mb of reads, the average coverage is 250 ÷ 5 = 50×.

Bacterial genomes range from under 2 Mb to more than 10 Mb, so the same amount of data gives very different coverage depending on the organism. The table shows how much data different genome sizes need, assuming 2 × 150 bp paired-end reads.

How much is enough?

Desai and colleagues assembled real Illumina data from E. coli (4.6 Mb), a yeast and a nematode at different depths with five assemblers. For most assemblers, 50× was the optimum read depth; one assembler needed 100× ^[1]. Assemblers have improved since then, and SPAdes is now among the most widely used for bacterial genomes ^[2], but 50–100× of good-quality data after trimming remains a practical target for a short-read bacterial assembly.

More data is not always better. Beyond about 100×, extra depth adds little to a short-read assembly, and very deep data are often downsampled before assembly. Low coverage causes more problems: gaps, more contigs and missed genes, which can matter when you are screening for AMR or virulence genes.

Read quality matters as much as depth

FASTQ files store each read together with a quality score for every base ^[3]. Before assembly, tools such as fastp trim adapters, remove low-quality bases and filter poor reads in a single pass ^[4]. This step reduces the amount of usable data, so a run that looks like 60× raw may give 50× after trimming. Contamination with another organism also lowers the useful coverage for your target genome, which is why read QC always comes before assembly.

What to send for analysis

• Raw FASTQ files, compressed as .fastq.gz. For paired-end data, send both files (R1 and R2) for every sample.
• Untrimmed reads. Trimming is part of the analysis, and doing it twice can remove useful data.
• The sequencing platform, read length and library kit, if you know them.
• Strain names and the expected species, so results can be checked against expectations.
• Long reads (Oxford Nanopore or PacBio) as FASTQ files too, if you have them. See choosing an assembly strategy.
• If you only have an assembled genome (FASTA), it can still be annotated and analysed, but read quality and coverage can no longer be checked.

How coverage appears in a BacSeq report

Every BacSeq Analytics package starts with read QC using fastp and assembly with SPAdes. The assembly QC section reports average coverage alongside genome size, contig count and N50, which are explained in assembly quality metrics. If coverage is too low for reliable results, we tell you before the analysis goes further. You can upload FASTQ, FASTA or contig files directly.