Segmentation fault

[jaredbernard]

Hi there, Robert,
This problem may be easy to solve, but apparently not for me. While trying -cluster_fast, I get the following error:

usearch12/bin/usearch12 -cluster_fast trimmed_codons_GR2.fasta -sort length -id 0.75 -centroids GR_centroids.fasta -uc GR_clusters.uc
usearch v12.0, 16.6Gb RAM, 8 cores
(C) Copyright 2013-24 Robert C. Edgar.
https://drive5.com/usearch

cluster_fast trimmed_codons_GR2.fasta
00:00 1.1Mb .. ...  (working)Segmentation fault (core dumped)

Any ideas?
Thanks!

[rcedgar]

At a guess, you are trying to cluster sequences which usearch is not designed to handle, what are they & what is the maximum sequence length?

[jaredbernard]

Thanks so much for replying. These are genes for chemosensory proteins gathered from several genomes. They have been trimmed, so they have a max length of 138 bp. I wanted to use uclust to weed out any that are too divergent.

[rcedgar]

Ok, they should work, no idea -- if you make a toy example per instructions here https://drive5.com/usearch/toy.html I'll take a look

[rcedgar]

Works for me under WSL. Please check if you get the error using the "official" binary:

wget https://github.com/rcedgar/usearch12/releases/download/v12.0-beta1/usearch_linux_x86_12.0-beta
./usearch_linux_x86_12.0-beta -cluster_fast toy.fasta -id 0.75 -centroids GR_centroids.fasta -uc GR_clusters.uc -threads 8

[jaredbernard]

That worked! Thanks so much! By the way, does the -sort option work with -cluster_fast usearch12? I ask because I want to use uclust (-cluster_fast being the main command to do it) but the newest version doesn't appear to have the -sortbylength command. I just want to be sure I'm giving uclust the best input.

…

On Wed, Dec 18, 2024 at 5:35 PM Robert Edgar ***@***.***> wrote: Works for me under WSL. Please check if you get the error using the "official" binary: wget https://github.com/rcedgar/usearch12/releases/download/v12.0-beta1/usearch_linux_x86_12.0-beta <https://urldefense.com/v3/__https://github.com/rcedgar/usearch12/releases/download/v12.0-beta1/usearch_linux_x86_12.0-beta__;!!PvDODwlR4mBZyAb0!TsQWY6Kyat-7qfM_oyExTBHzeXbyAje2Wj4CRt6th1UW9afjYlrFi8vNqrz4jq9JcWAMuobB52YNoGqhh4lMvoMteA$> ./usearch_linux_x86_12.0-beta -cluster_fast toy.fasta -id 0.75 -centroids GR_centroids.fasta -uc GR_clusters.uc -threads 8 — Reply to this email directly, view it on GitHub <https://urldefense.com/v3/__https://github.com/rcedgar/usearch12/issues/15*issuecomment-2552532252__;Iw!!PvDODwlR4mBZyAb0!TsQWY6Kyat-7qfM_oyExTBHzeXbyAje2Wj4CRt6th1UW9afjYlrFi8vNqrz4jq9JcWAMuobB52YNoGqhh4k8o8HliQ$>, or unsubscribe <https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AISSNL6YJA6GJCKJDTO4DI32GIIFXAVCNFSM6AAAAABT3R665GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNJSGUZTEMRVGI__;!!PvDODwlR4mBZyAb0!TsQWY6Kyat-7qfM_oyExTBHzeXbyAje2Wj4CRt6th1UW9afjYlrFi8vNqrz4jq9JcWAMuobB52YNoGqhh4mfgaXYEA$> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[rcedgar]

So where did the binary come from if not the github release? Something appears to be wrong with the build.

[rcedgar]

Correction: you're right, I forgot sortbylength is not in v12. A command-line option works unless you get an error or a warning that the option is not used.,

[jaredbernard]

I had followed these steps:

git clone https://github.com/rcedgar/usearch12
cd usearch12/src
make

I had also tried usearch11. I don't understand why, but the binary you shared with me worked! It must not be identical to the one available via cloning the directory. Thanks again, Robert!

[rcedgar]

*correction -- so I guess there is a problem with the Makefile, not sure what is going on there...