FusionDirect needs a BED file containing four columns (chro-
mosome, start position, end position, gene name). If this file is not
provided, FusionDirect will use the built-in BED file, which con-
tains most fusion genes of high clinical importance.
FusionDirect is available at:https://github.com/OpenGene/
FusionDirect.jl. It is written in Julia, which is a fresh language
allowing high performance technical computing. FusionDirect is
built upon the OpenGene Julia library (https://github.com/
OpenGene/OpenGene.jl), which provides basic sequence and var-
iant representations and I/O functions of regular NGS-related file
formats (i.e., FASTQ/FastA/VCF).2.5 Deduplication
and Unique Supporting
Read Counting
When it comes to determine the confidence of a called variant, the
most important evidence is the number and quality of its support-
ing reads. To calculate numbers of supporting reads, we need to
identify and collapse duplicated reads.Fig. 6FusionDirect result example. In the result, an EML4-ALK fusion is detected and reported with three
supporting read pairs, while two of them are unique. The reads of each pair are overlapped so they are merged
by pair before detection applied
Bioinformatics Analysis for Cell-Free Tumor DNA Sequencing Data 83