How do I calculate the probability of sequence coverage at particular window?

[vkpilla]

Hi all,

A newbie question:

How do I calculate the probability of a random set of sequences (at a specified length, say short reads of 25bp) aligning to a set window length (say 10kb)? Essentially, I'd like to know the sequence coverage probability along a specified length of DNA.

I'd like to use this sequence coverage probability to test whether what I see (for example, say I see 3 reads within a particular 10kb window) is truly significant or aligned by random chance.

Please let me know your thoughts and whether this is a valid question to ask in the first place.

Thanks!

[Simon Anders]

If I understand your question correctly, you want to calculate the probability that a read with a completely random sequence aligns to a place in your genome by chance. The answer is usually: zero.

There are 4^25=1.1e15 possible 25-bp reads. The human genome has 3e10 base pairs, hence the probability of a random 25-mer actually occurring in the human genome is roughly 3e10/1e15=2e-6.

Hence, if you see a read aligning somewhere, it has almost certainly been amplified from a real biological template, i.e., it is either from the sample or from contamination.

Contrary to popular belief, there is no such thing as alignment noise in high-throughput sequencing.

[Fad2012]

Hello guys

I am having almost the same problem, and i am confused of how to calculate the probability, here is my issue:

I have sequenced a PCR fragment of 2kb from original reference sequence of 7.5kb. I used illumina HiSeq paired-ends technology to generate 5 million 80bp reads with a coverage of x30 as I am looking for a recombination event between two serotypes of viruses, which is rare event.

I know that the event occurs by 1%, so I expect to find 1% of the reads represent the recombination event. Among these reads there are some reads which are going to span the junction point of the Recombinants, and therefore not aligned to any of the reference sequences. I want to calculate the probability of the coverage of those reads which span the junction point to calculate the error rate between the expected and the observed.

Need help!

Many thanks

[Fad2012]

Hi again

I forgot to add that the junction point could be occur at any nucleotide lies in the 2kb fragment.

Thanks a lot