'Error in generated code:', file=sys.stderr

Dear SALSA programmers,

I am trying to run SALSA in a bioconda-installed version. Specifically, I installed it as follows:

mamba create -n salsa2_2.3 salsa2=2.3 ;

I then prepared a sorted *.bed file and length file as instructed:

conda activate bedtools_2.30.0 ;
bamToBed -i $PROJECT/necator/Hi-C/2021.12.16/N_amer_2021.10.26.OmniC.dedup.mapped.PT.bam > N_amer_2021.10.26.OmniC.dedup.mapped.PT.orig.bed ;
sort -k 4 N_amer_2021.10.26.OmniC.dedup.mapped.PT.orig.bed > N_amer_2021.10.26.OmniC.dedup.mapped.PT.bed ;
conda deactivate ;

conda activate samtools_1.14 ;
samtools faidx N_amer_2021.10.26.fa ;
conda deactivate ;

Finally, I tried running SALSA:

conda activate salsa2_2.3 ;
run_pipeline.py \
--assembly N_amer_2021.10.26.fa \
--length N_amer_2021.10.26.fa.fai \
--bed N_amer_2021.10.26.OmniC.dedup.mapped.PT.bed \
--enzyme DNASE \
--output $PROJECT/necator/Hi-C/2021.12.27.01/salsa \
--cutoff 10000 \
--iter 3 \
--dup N_amer_2021.10.26.salsa_dups.txt \
--exp 234000000 --clean yes --prnt yes ;
conda deactivate ;

SALSA2 ran for a short while, but then hit the wall and died, with the following error messages:

bedfile loaded
Starting Iteration 1
bedfile started
bedfile loaded
Traceback (most recent call last):
  File "/ocean/projects/mcb190015p/schwarze/anaconda3/envs/salsa2_2.3/bin/fast_scaled_scores.py", line 2, in <module>
    import networkx as nx
  File "/ocean/projects/mcb190015p/schwarze/anaconda3/envs/salsa2_2.3/lib/python2.7/site-packages/networkx/__init__.py", line 68, in <module>
    import networkx.utils
  File "/ocean/projects/mcb190015p/schwarze/anaconda3/envs/salsa2_2.3/lib/python2.7/site-packages/networkx/utils/__init__.py", line 2, in <module>
    from networkx.utils.decorators import *
  File "/ocean/projects/mcb190015p/schwarze/anaconda3/envs/salsa2_2.3/lib/python2.7/site-packages/networkx/utils/decorators.py", line 7, in <module>
    from decorator import decorator
  File "/ocean/projects/mcb190015p/schwarze/anaconda3/envs/salsa2_2.3/lib/python2.7/site-packages/decorator.py", line 162
    print('Error in generated code:', file=sys.stderr)
                                          ^
SyntaxError: invalid syntax

python2 /ocean/projects/mcb190015p/schwarze/anaconda3/envs/salsa2_2.3/bin/RE_sites.py -a /ocean/projects/mcb190015p/schwarze/necator/Hi-C/2021.12.27.01/salsa/assembly.cleaned.fasta -e DNASE > /ocean/projects/mcb190015p/schwarze/necator/Hi-C/2021.12.27.01/salsa/re_counts_iteration_1
python2 /ocean/projects/mcb190015p/schwarze/anaconda3/envs/salsa2_2.3/bin/make_links.py -b /ocean/projects/mcb190015p/schwarze/necator/Hi-C/2021.12.27.01/salsa/alignment_iteration_1.bed -d /ocean/projects/mcb190015p/schwarze/necator/Hi-C/2021.12.27.01/salsa -i 1 -x N_amer_2021.10.26.salsa_dups.txt
python2 /ocean/projects/mcb190015p/schwarze/anaconda3/envs/salsa2_2.3/bin/fast_scaled_scores.py -d /ocean/projects/mcb190015p/schwarze/necator/Hi-C/2021.12.27.01/salsa -i 1

Why should a bioconda package for SALSA2 running in its own versioned environment have this problem? How should I fix it?

[SchwarzEM]

Thanks for the pointer about conflicting networkx versions. I have devised a method for using conda and pip to create an environment within which I have successfully run SALSA2. This may be useful to users who find conda to be a useful tool for managing versions of Python and other software. Here is what I did which worked:

mamba create -n salsa2_reqs python=2.7.14 ;
conda activate salsa2_reqs ;
pip install networkx==1.11 ; 

cd $PROJECT/src ;
git clone https://github.com/marbl/SALSA.git SALSA2 ;

cd $PROJECT/src/SALSA2 ;
make ;
mkdir bin ;
mv -i *.sh *.py break_contigs break_contigs_start correct_links bin ;
conda deactivate ;

[move to working directory, and run the following as a 1-CPU batch job:]

conda activate salsa2_reqs ;
PATH="$PROJECT/src/SALSA2/bin:$PATH" ;
export PATH ;
run_pipeline.py --assembly N_amer_2021.10.26.fa --length N_amer_2021.10.26.fa.fai \
--bed N_amer_2021.10.26.OmniC.dedup.mapped.PT.bed --enzyme DNASE --output $PROJECT/necator/Hi-C/2021.12.27.01/salsa \
--cutoff 10000 --dup N_amer_2021.10.26.salsa_dups.txt \
--exp 234000000 --clean yes --prnt yes ;
conda deactivate ;

… which worked! No crash at the networkx-invoking step; full run to near-chromosomal scaffolds.

I think the key step is using conda to create an invokable environment that has nothing but python 2.7, and then using pip to customize that environment by installing networkx 1.11 while actively in that environment, which will then be the sole active networkx version whenever that environment is reactivated. That works and allows SALSA2 to function; nothing else seems to.

[Surbhigrewal]

Thank you for this solution. Pip install networkx instead of conda install networkx in salsa environment solved the problem.