Arriba and STARfuison align

[xiucz]

Hi, The picture below illustrates the differences between the alignment results of Arriba (v2.4.0, STAR2.7.8a) and Starfusion (v1.10.0, STAR2.7.8a). In Starfusion's BAM file, there are numerous softclip reads that support the TIMM23--LINC00843 fusion events (chr10:51606988--chr10:51732772). It appears that the parameter settings used by Arriba's STAR alignment discarded many of these reads. Could you provide me with some suggestions?

arriba code(use the raw run_arriba.sh script)

arriba_v2.4.0//run_arriba.sh ~/arriba_v2.4.0//database//hg19_GENCODE37lift37/STAR_index_hg19_GENCODE37lift37/ ~/arriba_v2.4.0//database//hg19_GENCODE37lift37/GENCODE37lift37.gtf ~/arriba_v2.4.0//database//hg19_GENCODE37lift37/hg19.fa ~/arriba_v2.4.0//database//blacklist_hg19_hs37d5_GRCh37_v2.4.0.tsv ~/database//known_fusions_hg19_hs37d5_GRCh37_v2.4.0.tsv ~/arriba_v2.4.0//database//protein_domains_hg19_hs37d5_GRCh37_v2.4.0.gff3 6 ~/trim/A2_1.trim.fastq.gz ~/trim/A2_2.trim.fastq.gz

STARfusion code

~/STAR-Fusion-v1.10.0/STAR-Fusion --left_fq ~/trim/A2_1.trim.fastq.gz --right_fq ~/trim/A2_2.trim.fastq.gz --genome_lib_dir ~/hg19_GENCODE37lift37/STARFusion/GRCh37_gencode_v19_CTAT_lib_Mar012021.source/ctat_genome_lib_build_dir --CPU 6 --STAR_PATH ~/STAR-2.7.8a/bin/Linux_x86_64_static/STAR --examine_coding_effect --tmpdir  ~/tmp/ --output_dir STARfusion

Best, xiucz