Need help for “Strength”

[Masterchiefm]

Hello, I'm using the predict_from_custom_sequence function. The program works and gave me excellent output.

But I'm confused about the "strength" of the plot. each site has a strength score, but which one should be considered to be a "splice site"? can you give me a threshold？

as you can see, some of them are higher than 0.1, but the plot looks flat. Can you tell me how to set the threshold? I read your paper and SpliSER, but I cannot get a certern number.

您好，请问如何确认预测位点的threshold阈值呢？我的输出里有的位点平均都低于0.1，但是图片看起来很波动，有的图看着平坦，但是有些位点高于0.1. 我看了您的文章与 SpliSER， 但我并不能得到确切的阈值数值。请问strength分数有推荐的设置吗？

[SilverNebula]

Hello,
Since all analyses in our manuscript were based on Δscores for SNVs, we did not formally establish a threshold for raw splice site predictions.
If you are interested in identifying potential splice sites in a given sequence, we recommend using a threshold of 0.25 as a starting point. This value approximates the threshold generated by "top-k accuracy" in our test dataset