GASOLINE

I would like to commend your innovative approach to small-scale somatic structural variant (SV) detection, particularly the implementation of the Normalized Reciprocal Overlap (NRO) dynamic SV signal merging. However, During my use of GASOLINE, I encountered several issues that required modifications for proper functionality. Below are the problems I encountered: 1. In GASOLINEParse.pl (line 66), the genome selection statement only included the HG37 genome, ignoring the HG38 genome. "#### Checking Bed File ### if (FileBed4Sam eq ""){ if (Assembly eq "hg19"){ FileBed4Sam = "$ToolPath/DB/ChromosomeCoordinate_HG37.bed"; } if (Assembly eq "hg38"){ FileBed4Sam = "$ToolPath/DB/ChromosomeCoordinate_HG37.bed"; } }" 2. In the "Lib/R/SomaticConvert2VCFFunction.R script" (line 214), the code for reading tables: "TableGASOLINEInterTransIn <- read.table(FileOutGASOLINEInterTransSomaticCall, header = FALSE, sep =" ", quote="") " to TableGASOLINEInterTransIn <- read.table(FileOutGASOLINEInterTransSomaticCall, header = FALSE, sep =" ", quote="", fill = TRUE, check.names = FALSE). The original format of the FileOutGASOLINEInterTransSomaticCall file contained rows with fewer than 20 columns, causing the original code to fail. Applying this code to "TableGASOLINEInterTransIn <- read.table(FileOutGASOLINEInterTransSomaticCall, header = FALSE, sep =" ", quote="", fill = TRUE, check.names = FALSE)." , the script ran successfully. 3. In the GASOLINESomaticVCFCreate.pl script, the original script did not include contig-related information when generating the VCF format. This omission can cause issues in downstream processing of the VCF file. For future improvements, it would be beneficial to further enhance the VCF file by including additional annotation information to ensure comprehensive data representation. 4. In the GASOLINESomaticVCFCreate.pl script, Line 214, when determining the SV types, the BND type is separated from the other four types (INS, INV, DEL, DUP), and generation of two somatic txt files: .Results.Somatic.txt (for INS, INV, DEL, DUP) and Results.InterTrans.Somatic.txt (for BND). These two txt files are later merged to output the final somatic VCF file. However, there is no check for the existence of the BND file, leading to an error when running the program if no BND SVs are detected. After adding the corresponding if statement, the program can run as expected. These modifications were validated in my local implementation and resolved the observed issues. I hope that thees issues and solutions can help you further improve and refine GASOLINE.

Hi, I read your paper from biorx. I have not used the tool but I have some questions. Why does it take 3 hours for the tool to parse the Bam file? To detect somatic SVs couldn't you just use Sniffles2 or cuteSV(v2.0) to find the structural variants in a control sample, do the same thing with the cancer sample, and then compare the calls with the p-score like you did. I have used both Sniffles2 and cuteSV on about ~10x coverage ONT and they finish calling in like 5 minutes with 10cpu and 120G ram. I understand your clustering methods lead to better grouping and detection of small SVs. Is that the main focus of the tool? Why would I use this tool if it takes waayy longer than the competition with slightly better results? Are there any plans to improve the efficiency of the tool in the future? ALSO could you add in functionality where it does the somatic determination from 2 VCF files rather than having to both VCF files from scratch?