CNV detection with single-read data (bacterial genome)

[MarinePAM]

I am planning to do a CNV calling for the first time and I've got Illumina single-read data (50bp). Libraries were not sequenced in paired-end because it is a bacterial genome so the coverage is very high even with single-read data (~400X).

I've read a lot of things about CNV calling and discovered that most of tools are designed for paired-end data.
I've read that there are different kind of methods depending on what the tool focus on : SR (Split Read : split misaligned paired reads), RP (Read Paired : uses insert sizes distribution), RD (Read Depth) and CA (Combinatory approach : takes advantages of all methods). The latest is the more confident and complete method, so that I would like to use a tool which use the CA method. But it seems that paired-end reads are always required and maybe I have no other choice than using a RD tool.

My question is: did someone already call CNV with single-read data, or does someone know which tool might be the best suited for?

[BnaT]

There are a few algorithms available, I'd try Socrates (split-read) which works either on PE or SE data.

http://bioinf.wehi.edu.au/socrates/

[MarinePAM]

Hi BnaT, thank you a lot for your answer ! Sorry I did not see it earlier !
I'm glad to hear about a tool I did not know, I will have a look. So far I tried Control-FREEC and CNV-seq, both using the read depth to compute CNV.

[MarinePAM]

A personal reflexion which may interest someone else : I have 50nt single-end reads, maybe a method based on splitting read is not the more reliable for such small reads (because an alignment of a few bases may be not specific enough)