tjiangHIT / cuteSV

Long read based human genomic structural variation detection with cuteSV
MIT License
243 stars 36 forks source link

cuteSV with ONT amplicon data #148

Open ctarpey opened 1 month ago

ctarpey commented 1 month ago

Hello!

I have ONT data of a single amplicon that is ~4500 bp in length and would like to identity SVs. I am interested in any SVs present, but specifically, there may be repeats introduced by a hairpin structure that is ~22bp in length. Do you have any suggestions for parameters that I should change from the suggested ONT defaults to increase the accuracy and sensitivity of the programs calls in such a short region?

It may be radical, but I was considering changing: --max_cluster_bias_INS to 30, --max_cluster_bias_DEL to 30 --max_cluster_bias_INV to 50 --max_cluster_bias_DUP to 100 --min_size to 22 --remain_reads_ratio to 0.95

Am I off base in changing these parameters, and are there others that I should consider?

Thank you!

Carolyn

Meltpinkg commented 1 month ago

Hello, @ctarpey

cuteSV is dedicated to detecting SVs of over 50bp, while also obtaining the ability to detect smaller variations. Though I haven't seen the specific form of your input data, in my thought, it can be successfully detected. The parameters that you changed mainly decrease the tolerance of the errors or deviation in sequencing or mapping. It will work well when your alignment file has a high degree of neatness. In my suggestion, the parameters can be adjusted according to the quality of your ONT data to achieve the better performance, and maybe you can have some try since I cannot give an accurate answer for diverse data. Besides the parameters mentioned above, another parameter which may influence the detection is the -s. You can adjust this parameter according to the coverage of the sequencing.

Hope it will help!

Best, Shuqi