x coverage (or -fold covergae is used to describe the sequencing depth. For example, if your genome has a size of 10 Mbp and you have 100 Mbp of sequencin data that is assembled to said 10 Mbp genome, you have 10x coverage.
% coverage is used, e.g., in sequence mapping. If you have a reference genome of 10 Mbp and 70 Mbp sequence data from a different strain mapping to 7 Mbp of the reference, you have 70% coverage of the reference (with a 10x coverage of the covered regions).
x coverage (or -fold covergae is used to describe the sequencing depth. For example, if your genome has a size of 10 Mbp and you have 100 Mbp of sequencin data that is assembled to said 10 Mbp genome, you have 10x coverage.
% coverage is used, e.g., in sequence mapping. If you have a reference genome of 10 Mbp and 70 Mbp sequence data from a different strain mapping to 7 Mbp of the reference, you have 70% coverage of the reference (with a 10x coverage of the covered regions).
Thank you very much. So, what x-coverage better gives 100% coverage (if any)?
For de novo assemblies that depends on a number of factors: G+C content, sequencing technology, and the library preparation method to name just a few.
50x coverage from a MiSeq run using a PCR-free library is usually a very good start. I work with Actinomycetales and assemblies with that amount of data usually give >99.99% coverage of the finished genome. For example, in a 3 Mbp genome I usually do not get < 100 bp, depending on the G+C content (which often exceeds 72%).
For mappings, the same holds true, given that the source of the data and the reference are similar enough.
The word "coverage" has always been used loosely and therefore has several different meanings. First, it can represent the same concept as the words "fold", "depth", and "redundancy", which is the number of times the *average* base position is spanned by data (a read, a clone, etc.) Similar to Christian's example above, 40Mb of sequence data for a 4Mb bacterial genome would be 10X or 10-fold "coverage". Conversely, "coverage" can also mean the fraction of a target that is represented in the data according to some threshold. That same 40Mb may not, in fact, represent the *entire* bacterial genome. We might say, for example, that only 99.9% of the genome is "covered". Various mathematical models (some of which I have proposed) link these 2 concepts together, for example X amount of data depth gets you a fraction Y of the genome. Wikipedia has a good further discussion of this problem at https://en.wikipedia.org/wiki/DNA_sequencing_theory
Great, Thank you guys!
How do you use the coverage to see the difference at strain level.
What is the maximum coverage a new strain should have as compared to the reference genome?
@Tsegaye, that's an impossible question to answer as it depends very heavily on the species in question and the expected relationship between the new strain and the reference. Look at the early E. coli genome papers for some examples of massive differences in genome content between strains.
Dear all,
The coverage should be consider as haploid genome size or diploid genome size for a typical diploid organism
Gopal Singh it is usual to report X coverage relative to the *haploid* genome size. It is not always explicitly stated when recommendations are given (e.g. when assembling a diploid genome), so it's generally best to check when it's not clear. It also varies when reporting empirical read depths from read mapping, e.g. if you have mapped reads onto a phased diploid genome reference, all of your read depths will be diploid read depths. As with most things, being explicit is the best policy to avoid possible confusion or errors.
Hello Richard,
Thank you for these valuable explanation.
One of the major challenge in genome assembly specially de novo to generate the scaffolds (based on short reads and/or long reads) compatible to Hi-C data for a phased genome. In this context, we have to generate sufficient reads to fulfill genome-wide coverage.
Kindly correct me if I am leading to some wrong understanding.
Gopal Singh I've not used Hi-C myself. If you are unsure of coverage requirements, it's probably best to look at papers where it has been done successfully and work out whether they are talking about haploid or diploid sequencing coverage.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology