Àngèlo Dúárte̊

Thanks for your prompt answer. I will try your suggestion and check if the results match. Meanwhile, please consider our help in implementing the gap extension penalty in Parasail.

[gotoh.pdf](https://github.com/jeffdaily/parasail/files/739002/gotoh.pdf) To use open+extend generates wrong results. The extend penalty (b) should be multiplied by the gap length (k) before adding to the gap opening penalty (a). The final formula...

The bioinformaticians from the team have compared the results of both, the final alignment string and the obtained scores, using the alignment tool provided by NCBI, which uses the Needleman...

We are studying the best way to explain the problem to you. Since our team has computer scientists and bioinformaticians, sometimes we have to synchronize the ideas. I'll contact you...

Do not worry. We are on the same side.

We read the Parasail nw.c source code and found that the difference in the results is caused by the way Parasail calculates the gap penalty and does the matrix initialization....

> > Hi, after downloading your latest model "uniref_embeddings", I found that the dimensionality of the protein word vector is not 100 dimensions, it is greater than 900 dimensions, so....