evolbioinf / andi Goto Github PK

Hello Fabian,

Andi is great. So powerful. I have two genomes and I want to see very fast in which positions in the genome differ.
Can I get this with andi?

Fabian

I've tried to do it : https://github.com/Homebrew/homebrew-science/pull/3305
But having problems with divsufsort. It seems to find it, but something isn't right.
Does it use pkgconfig to discover it?
It seems to work on Linux but not on Mac.

checking divsufsort.h usability... no
checking divsufsort.h presence... yes
configure: WARNING: divsufsort.h: present but cannot be compiled
configure: WARNING: divsufsort.h:     check for missing prerequisite headers?
configure: WARNING: divsufsort.h: see the Autoconf documentation
configure: WARNING: divsufsort.h:     section "Present But Cannot Be Compiled"
configure: WARNING: divsufsort.h: proceeding with the compiler's result
checking for divsufsort.h... no
checking for divsufsort in -ldivsufsort... yes
configure: error: Configuration for libdivsufsort failed. Either install libdivsufsort, or use our replacement, psufsort, instead.
    ./configure --without-libdivsufsort
The latter may result in longer runtimes.

If I disable I still get error here: #6

./configure not actually present in repo. autoconf didn't run, maybe?

Hi,

would it be possible to also have statically linked binaries available for download?

That would help to have the newest version included in bioconda, since there is a problem with the gsl version requirement on their build environment.
In this case, it would be easier to have the official static binary used by bioconda.

thanks!

The PHYLIP format for distance matrices only allows identifiers of up to 9 (or 10?) characters. Unfortunately this means that sometimes names are cutoff, making them indistinguishable. A --name-length option might help.

I compared 40668 viral genome sequences, however, the dimension of the ouput distance matrix is 40668, 81338. Software prompts that some sequences cannot be calculated, but it is puzzling that the output distance matrix is asymmetric.

After some experiments at home I realised we have to improve the autoconf detection:

• If libdivsufsort is not installed, thrown an error
• Also check for the divsufsort.h header
• dito for GSL

This occurs on OSX compiler with both GCC and CLANG:

clang -DHAVE_CONFIG_H -I. -I../src   -I/usr/local/opt/libdivsufsort/include -I/usr/local/Cellar/gsl/1.16/include  -g -O2 -c -o reallocarray.o reallocarray.c
strchrnul.c:7:14: error: use of undeclared identifier 'NULL'
        return p != NULL ? p : strchr(s, '\0');

I'm guessing NULL is defined in stdio.h or stdlib.h and you have no #include in that C file.

The tests for low homology, nan, and join are not reproducible yet, leading to build failures in Debian and Ubuntu. Use the fixed seed as for the other tests.

Can we link against the libdivsufsort from here?
https://github.com/simongog/sdsl-lite

Hello

I have been trying to run this on a set of Pseudomonas genomes (6000). Some of them are draft genomes so I needed to run the option --join.
Not even 10 minutes after starting the command I got the error:
zsh: killed andi --join *.fna > matrix.mat

I have to say I do not have access to a cluster and I do most of the work on a Macbook with M1 chip. I have tried the above command with a small set of genomes and it worked so the installation is right. Is the error due to computational power?

Expected behaviour: a warning should be printed.

FYI - I managed to get it packaged:

https://github.com/Homebrew/homebrew-science/pull/2372

minAnchorLength makes the anchors longer than necessary. Fix that.