blastn: 2.2.31+ Package: blast 2.2.31, build Jun 2 2015 10:47:46

Question

[Luca]

Hi Everyone, I'm trying to install on my ubuntu docker blastn

blastn: 2.2.31+
Package: blast 2.2.31, build Jun  2 2015 10:47:46

The main problem is that I do not find anywhere the binary or a way to install it. Unfortunately with all the blast+ that are out there today I cannot reproduce an analysis so I really need this one.

Thanks!

Answer 1

[GenoMax]

You can find the binaries for 2.2.31 here: https://ftp.ncbi.nih.gov/blast/executables/blast+/2.2.31/

That said if you can't reproduce an analysis with a different version/newer version that result would be suspect.

[Michael]

Try this one: https://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.2.31/ (it's now ncbi.nlm.nih.gov)

[Luca]

I agree, the main problem is that the command line is hidden in a jar file, and this jar file uses blastn. My idea is that for some reasons some parameters are slightly different between the versions... Here is the comparison in a plot

That's too different.

[SequenceServer]

Ugh that's frustrating. Are you using the same version of Linux and all data files too? BLAST has some dependencies that may have also changed

[Luca]

yeah it is in a docker environment and the only difference is in the blast version.... that's crazy, installing the suggested version I have almost everything blue. So is really just the blast version... but can it give such different results?? this is crazy

[SequenceServer]

Yes...

BLAST, like many other bioinformatics algorithms, is a heuristic. This means it's not "a perfect search" but instead takes many algorithmic shortcuts that try to appropriately trade off so that in a reasonable amount of time we get the result we want. They are constantly improving the algorithm to improve runtimes and memory usage, including as database sizes continue to increase. Most other bioinformatics tools have similar things going on. As a biologist it's important to remember that the strong signals should be robust to choice of algorithm. But as you can see, for many edge cases there can be some ambiguity.

Something you can look at are the -task options - where for each algorithm, multiple different options now exist that are optimised for different tasks.