FEM is a Fast and Efficient short read Mapper. Currently, FEM can return all mapping locations of NGS single-end short reads with up to 7 errors.
We pushed the source code of FEM purely in C to replace the old hybrid one. Note that the index construction becomes slower since qsort is slower than STL sort.
The source code of FEM is avaliable. We removed the dependence on Intel TBB library so that FEM can be easily used and tested. A parallel sorting without dependency on third party libraries will be added to FEM soon.
FEM index <window size> <step size> <reference.fa>
Usage: FEM align [options]
Options:
-e INT error threshold
-t INT number of threads
-f STR seeding algorithm: "g" for group seeding and "vl" for variable-length seeding
-a use one additional q-gram for filter
Input/output:
--ref STR Input reference file
--read STR Input read file
-o STR Output SAM file
To reduce mapping time, we recommend to use the smallest step size as long as the index can fit into the memory. In next version, FEM will choose the step size according to given memory adaptively.
If you use FEM, please cite Fast and efficient short read mapping based on a succinct hash index.
Haowen Zhang
Email: hwzhang@gatech.edu
We welcome any bug report and suggestion. Please start an issue in the repo.