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...

[hmontenegro]

I can confirm the official binary works, but compiling from source results in segmentation fault, even if running usearch with no parameters. This is on Rocky Linux 8, GCC 8.5.0 and ccache 3.7.7.

[rcedgar]

@hmontenegro can you compile with debug symbols and post gdb traceback?

[hmontenegro]

Adding some information, USEARCH compiles and runs fine (at least with small datasets) with newer GCCs: both GCC 11.2.1 on Rocky Linux 8 and GCC 14.2.0 on Debian testing.

If you still think it would be helpful, I will give it a shot and report back.

[hmontenegro]

Here it is. I have no experience in debugging, hope I did everything correctly. I've added -ggdb to CXXFLAGS, and commented out strip $(BINPATH).

$ gdb ../bin/usearch12 
GNU gdb (GDB) Red Hat Enterprise Linux 8.2-18.el8
Copyright (C) 2018 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "x86_64-redhat-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
    <http://www.gnu.org/software/gdb/documentation/>.

For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from ../bin/usearch12...done.
(gdb) run
Starting program: /home/hmontenegro/src/USEARCH/usearch12-12.0-beta1/bin/usearch12 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".

Program received signal SIGSEGV, Segmentation fault.
0x0000000000000000 in ?? ()

[OilArtLovely]

Dear Dr. @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.,

Would it be possible to add back the -sortbylength option? Some older software still uses it, and the newer releases generate errors because of its removal.

Thanks for your consideration.

[websandip]

Dear Dr. @rcedgar,

Is it possible to reintroduce the -sortbylength flag? Some pangenome software still relies on this option, so having it available again would be very helpful.

Thank you