Two parameters you choose
So, how much data should you choose? Depending on the study objective, the key parameters are:
01
Amount of data
in gigabytes or number of reads
02
Coverage
the average number of times each nucleotide in the genome will be read/sequenced
Amount of data
What is data output in sequencing and when is it used?
In sequencing, data are usually expressed in two forms:
01
Number of reads (e.g. 30 M reads, meaning 30 million reads)
Reads are DNA fragments generated by the sequencer. Depending on the technology, they may include 1×50 bp or 2×150 bp reads, for example, corresponding to a single-end read 50 nucleotides long or paired-end reads 150 nucleotides long each.
02
Amount of data in Gb/GB (e.g. 10 GB)
This is the total number of sequenced bases (nucleotides) - the longer the reads and the more reads generated, the greater the amount of data.
EXAMPLE · 30 M PAIRED-END READS, 2×150 BP
Number of reads × 2 × read length
BASE
30 000 000 × 2 × 150 = 9 000 000 000 nucleotides
RESULT
~9 GB of raw data
The final amount of clean data in GB also depends on data quality and read filtering.
Units and clean data
UNITS
Gb
gigabase, meaning one billion nucleotides
GB
gigabyte, a unit of computer storage
Gb ≈ GB
in sequencing practice, 1 Gb ≈ 1 GB in raw data files (FASTQ)
Raw vs clean data
Raw data comprise all reads generated directly by the sequencer. Clean data are reads after:
01
Adapter removal
02
Low-quality read filtering
Q-score < 20-30
03
Removal of excessively short fragments
Typical loss: 5-15% of data during quality control.
Example: from 9 Gb of raw data (raw reads), approximately 8-8.5 Gb of clean data (clean reads) will remain for analysis after filtration.
Recommendations by project type
The metric and recommended amount of data depend on the project type. For projects without a single reference, conventional coverage is not used; the number of reads is used instead.
PROJECT TYPE
MEASURE
RECOMMENDATION
number of reads per sample
~30 M reads for gene expression analysis (DEGs)
~50-100 M for low-expression genes, long non-coding RNA or species with a very large number of genes
~10 M reads for
RNA-seq bacteria
number of readsno single reference, so conventional "coverage" does not apply
5-20 M reads for low-complexity samples
≥ 30 M reads or more for environmental samples
number of readsno single reference, so conventional "coverage" does not apply
50 000 reads/tags per sample for basic identification
> 100 000 reads for highly accurate taxonomy
coverage
standard coverage is 30×
coverage
typically 100-150×
Coverage
Coverage is the average number of times each nucleotide in the genome has been read/sequenced.
EXAMPLE · HUMAN GENOME, 90 Gb OF DATA
Amount of data ÷ genome size
BASE
90 Gb ÷ 3 Gb
RESULT
30× coverage
Each nucleotide will be read 30 times on average. The amount of sequenced data can be calculated using the formula: number of reads × read length, e.g. 30M x 2 x 150.
How to select the amount of data
01
Define the project objective
Are you sequencing a genome, exome, transcriptome or microbiome? Each project type requires a different amount of data.
02
Consider sample quality
Fragmentation, degradation or low DNA input may require greater coverage (even +20-30% additional data).
03
Define additional factors that may affect the required amount of data
The more demanding the analysis (e.g. detection of low-frequency somatic variants), the greater the required coverage. In addition, if the results will be compared with another project that has already been completed, maintaining the same service specification will make the results easier to compare.
Impact on project price
Yes, the amount of data has a major impact on the service price. Why?
01
More data = more sequencer run time
02
More data = greater reagent consumption
03
More data = longer bioinformatics analysis and larger files to store
More is not always better. It is important to find the optimal balance, taking into account the project type, its objective and the available budget.
Summary
Golden rules
01
More is not always better
02
Select parameters appropriate for the specific project objective
03
Consider sample quality
04
Allow a margin for losses during quality control
Choosing the amount of data requires a balance between result quality, the study objective and the budget.
Start by completing our intelligent form, which will guide you through a needs assessment. If you encounter any uncertainty, indicate in the form that you need assistance and send it to us. This will give us an overview of your requirements and allow us to help you during a consultation.
