custom sequence annotation about spliceai HOT 4 CLOSED

The second column in the output corresponds to acceptor probabilities for each position, and the third corresponds to donor probabilities. In this case, the model is saying that the input sequence doesn't splice.

from spliceai.

how would an example w/ a splice site look like? which scores would you expect (the ones in the 2nd/ 3rd column seem "small" already ...)

from spliceai.

Suppose we use this sequence 'CGATCTGACGTGGGTGTAGGTAAGTGCATTATCGATATTGCAT' (I have planted a donor motif around the middle), you get this output:

array([[[9.99985993e-01, 1.40094662e-05, 7.43200417e-08],
[9.99981284e-01, 1.81920605e-05, 4.59646884e-07],
[9.99959588e-01, 4.04304556e-05, 9.72240031e-08],
[9.99997914e-01, 1.81026769e-06, 2.11512273e-07],
[9.99981105e-01, 1.87257538e-05, 1.84678086e-07],
[9.99997616e-01, 2.22162589e-06, 1.58002280e-07],
[9.99976337e-01, 2.32699349e-05, 3.86139988e-07],
[9.99968171e-01, 3.15743491e-05, 2.48222989e-07],
[9.99869466e-01, 2.05758897e-05, 1.09932516e-04],
[9.99990284e-01, 8.31942270e-06, 1.38772452e-06],
[9.99941647e-01, 5.70513548e-05, 1.33313620e-06],
[9.99996543e-01, 2.48335687e-06, 9.24738913e-07],
[9.99904454e-01, 3.52658608e-05, 6.02765722e-05],
[9.99976814e-01, 2.26476441e-05, 5.33687967e-07],
[9.98460770e-01, 3.49480324e-05, 1.50430272e-03],
[9.99996006e-01, 3.18197226e-06, 7.64729862e-07],
[9.99990106e-01, 8.76577269e-06, 1.17254172e-06],
[9.99990821e-01, 6.36481946e-06, 2.79536471e-06],
[6.46120489e-01, 8.37011263e-04, 3.53042454e-01],
[9.99608636e-01, 3.86872416e-04, 4.51662299e-06],
[9.99994636e-01, 3.63963477e-06, 1.67586336e-06],
[9.99996066e-01, 1.63950494e-06, 2.23818938e-06],
[9.99969482e-01, 5.67788311e-06, 2.48165161e-05],
[9.99998391e-01, 1.49879145e-06, 1.63973709e-07],
[9.99952912e-01, 4.61158597e-05, 1.05664662e-06],
[9.99996006e-01, 3.80371307e-06, 2.04130629e-07],
[9.99997735e-01, 2.14823740e-06, 1.33420997e-07],
[9.99997497e-01, 2.08130041e-06, 4.22661458e-07],
[9.99998868e-01, 1.14169552e-06, 4.58909710e-08],
[9.99998093e-01, 1.14678699e-06, 7.67544975e-07],
[9.99999642e-01, 2.47565538e-07, 7.65863177e-08],
[9.99999225e-01, 7.40523660e-07, 2.01556567e-08],
[9.99994457e-01, 5.47419040e-06, 5.00959274e-08],
[9.99996185e-01, 3.27182170e-06, 6.35605090e-07],
[9.99997258e-01, 2.72508942e-06, 3.42512507e-08],
[9.99999166e-01, 8.44390911e-07, 3.03123286e-08],
[9.99998093e-01, 1.89125274e-06, 1.63006373e-08],
[9.99999702e-01, 3.21699702e-07, 9.29854238e-09],
[9.99999821e-01, 2.41913938e-07, 1.15190559e-08],
[9.99999404e-01, 6.25196776e-07, 8.32780334e-09],
[9.99997973e-01, 2.00116369e-06, 1.19149473e-08],
[9.99995232e-01, 4.75768866e-06, 1.50998183e-08],
[9.99999642e-01, 3.18003885e-07, 2.90503799e-08]]], dtype=float32)

FYI, you can use the vectors acceptor_prob and donor_prob instead of y.

from spliceai.

perfect thanks a lot!

from spliceai.