Which C++ Libraries Are Best For Dealing With Fastq Files?

Question

[Jeremy Leipzig]

I would like to rewrite some perl scripts into something faster. I haven't written C++ since the Clinton administration. Granted I am not married to C++ per se but I would need something that benchmarks well.

Which C++ libraries are people using to deal with NGS data?

[Phis]

Which OS/CPU architecture do you need it for?

[Jeremy Leipzig]

RHEL5 x86_64 ..

Answer 1

[Michael]

I saw a SeqAn poster at ISMB last year. No experience with the library (nor C++) myself but they support the fastq format and they made the impression that they are quite competent.

SeqAn file formats

Answer 2

[User 59]

I wouldn't dream of doing this I admit, I tend to handle fastq files with applications other people develop.

However there is a FASTA/FASTQ c++ parser here:

http://lh3lh3.users.sourceforge.net/parsefastq.shtml which might serve as a base for what you want to do.

It's from Heng Li who also works on SAMtools, BWA and MAQ

[Pierre Lindenbaum]

bouuhhhh in http://lh3lh3.users.sourceforge.net/kseq.shtml ANY malloc should be checked against NULL (line 56 , 121 , 188 ...) :-(

[Jonathan Manning]

+1 - Highly recommended. That header supports compressed files, too, which speeds IO-bound processing. One caveat for C++ though - that header wants char and FILE, not C++ strings and iostreams, but that's easy enough to manage.

Answer 3

[Manuel]

If you work with NGS data and want to try SeqAn as Michael already suggested, have a look at this tutorial for importing read data. Also, their documentation has greatly improved recently.

Answer 4

[Luiz Irber]

Another option is SeqDB

Answer 5

[cartoonist]

Since I could not find any C++ library that meets my requirements, I re-write the kseq library (by @lh3) in C++ using templates, called kseq++. Here, I compared its performance with original kseq and SeqAn: https://github.com/cartoonist/kseqpp

SeqAn uses its own string class. If one does not use it, converting back to std::string is really expensive (3x slower on my workstation).