specific location coverage from average coverage?

[Fad2012]

Hello there,

I want to know if there is any way to calculate the coverage at a specific nucleotide across the amplicon using the average coverage, with the assumption that the reads are normally distributed across the ampliocn.

Say I have sequenced a 1.3kb amplicon, with read length 300 nts, and number of reads 10000. using Lander/Waterman equation (LN/G), the average coverage will be ~2300x. This tells us that every nucleotide was covered at least with 2300 reads. My question is that, can I estimate roughly the actual number of reads cover a certain nucleotide across the ampliocn, say nucleotide number 9, or any other nucleotide. Is it doable in a theoretical way without using tools like bedtools or samtools?

Many thanks

[Brian Bushnell]

You can't get a useful answer without mapping. Did you amplify some region then randomly shear it? If you're sequencing the amplicons directly, the coverage will all be in the same location. If you're sequencing the randomly sheared amplicons, there's no way to predict what the coverage will look like.

[Fad2012]

mmmm I see..yes this is the plan, to shear the amplicon.

What If I use a simulated data, that is perfectly normally distributed, can I use this average coverage to calculate a specific point coverage?

Many thanks

[Brian Bushnell]

Certainly, if you use synthetic data with a normal distribution, it's possible to estimate the coverage at any point along the amplicon from the average.

[Fad2012]

Thanks very much Brian, could you please tell me a key though of how this can be done? Do I have to find first the actual coverage for one point, and use it as a reference? Or I can use the coverage average along with the mean and standard deviation? In both ways, I am not sure how to carry out this process...Confused

Could please help me further.

Much appreciated

[Brian Bushnell]

Well, it's possible, once you know the constants, using the formula:

https://en.wikipedia.org/wiki/Normal_distribution

Using 300bp reads instead of 1bp reads would make it harder; you'd need to modify the formula somehow. But it's an approximation either way, and not one that I think would be useful at all in the real world, so I don't see much point in putting in any effort to solve it.

[Fad2012]

Many thanks