Bowtie error in snp calling pipeline

[snayfach]

Bowtie error in snp calling pipeline

[snayfach]

Could you try running the unit test (cd MIDAS/test; test_midas.py)? Could you also provide me the list of commands you ran?

Thanks,
Stephen

[ddomman]

Hi @snayfach,

Here is the output from the unit test:

apps/MIDAS/test$ python test_midas.py 
..F..F..
======================================================================
FAIL: test_help_text (__main__.MergeGenes)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "test_midas.py", line 92, in test_help_text
    self.assertTrue(sum(self.retcodes)==0, msg=error)
AssertionError: 

Failed to execute the command: merge_midas.py genes 

======================================================================
FAIL: test_help_text (__main__.RunGenes)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "test_midas.py", line 64, in test_help_text
    self.assertTrue(run(self.command)==0, msg=error)
AssertionError: 

Failed to execute the command: run_midas.py genes ./sample -1 ./test.fq.gz -n 100 --species_id 57955 

----------------------------------------------------------------------
Ran 8 tests in 455.129s

FAILED (failures=2)

These are the commands I ran:

run_midas.py species ERR866577_species -1 ERR866577_1.fastq.gz -2 ERR866577_2.fastq.gz -t 20
run_midas.py snps ERR866577_species --species_id 57327 -1 ERR866577_1.fastq.gz -2 ERR866577_2.fastq.gz -t 20
run_midas.py snps ERR866577_species  -1 ERR866577_1.fastq.gz -2 ERR866577_2.fastq.gz -t 20

Both snps commands failed with the above bowtie error.

Thanks for the help,
Daryl

[snayfach]

Hi Daryl,

It looks like the unit test is failing. Could you send me the output from running:
cd MIDAS/test
run_midas.py genes ./sample -1 ./test.fq.gz -n 100 --species_id 57955

[ddomman]

Hi Stephen,

Sorry for a mix up, I realized I didn't have my python ENV activated and our system pysam version is old, and that is why it was failing.

After I realized this, and activated the ENV, the test_midas.py runs fine.

apps/MIDAS/test$ python test_midas.py 
........
----------------------------------------------------------------------
Ran 8 tests in 506.385s

OK

Despite this, I'm still getting the original error bowtie error for the snps command (after species has worked fine):

Error encountered executing:
apps/MIDAS/bin/Linux/bowtie2-build ERR866577_test/snps/temp/genomes.fa ERR866577_test/snps/temp/genomes

Error message:
Warning: Empty fasta file: 'ERR866577_test/snps/temp/genomes.fa'
Warning: All fasta inputs were empty
Error: Encountered internal Bowtie 2 exception (#1)
Command: bowtie2-build --wrapper basic-0 ERR866577_test/snps/temp/genomes.fa ERR866577_test/snps/temp/genomes 

The odd thing is is that ERR866577_test/snps/temp/genomes.fa is actually not empty. It appears to be the correct fasta files.

Update:

It appears that directly entering the command works.

apps/MIDAS/bin/Linux/bowtie2-build --wrapper basic-0 ERR866577_snps/snps/temp/genomes.fa ERR866577_snps/snps/temp/genomes

Produces relevant genomes.*.bt2 files

[snayfach]

It's strange that the unit test works but you're getting an error on your own data. Could you double check that this works:
cd MIDAS/test
run_midas.py snps ./sample -1 ./test.fq.gz -n 100 --species_id 57955

Are you trying to run run_midas.py snps twice at the same time with the same output directory (unclear from reading your earlier post)? If so, then this is probably the issue.

[ddomman]

Test data works fine:

apps/MIDAS/test$ run_midas.py snps ./sample -1 ./test.fq.gz -n 100 --species_id 57955

MIDAS: Metagenomic Intra-species Diversity Analysis System
version 1.0.0; github.com/snayfach/MIDAS
Copyright (C) 2015-2016 Stephen Nayfach
Freely distributed under the GNU General Public License (GPLv3)

===========Parameters===========
Script: run_midas.py snps
Output directory: ./sample
Remove temporary files: False
Pipeline options:
  -build bowtie2 database of genomes
  -align reads to bowtie2 genome database
  -use samtools to generate pileups and call SNPs
Database options:
  -include specified species id(s): ['57955']
Read alignment options:
  -input reads (1st mate): ./test.fq.gz
  -input reads (2nd mate): None
  -alignment speed/sensitivity: very-sensitive
  -number of reads to use from input: 100
  -number of threads for database search: 1
SNP calling options:
  -minimum alignment percent identity: 94.0
  -minimum mapping quality score: 20
  -minimum base quality score: 30
  -minimum read quality score: 20
  -trim 0 base-pairs from read-tails

Building database of representative genomes
  total genomes: 1
  total contigs: 1
  total base-pairs: 5163301
  0.16 minutes
  0.12 Gb maximum memory

Mapping reads to representative genomes
  0.01 minutes
  0.12 Gb maximum memory

Running mpileup
  0.07 minutes
  0.12 Gb maximum memory

Formatting output
  3.14 minutes
  1.09 Gb maximum memory

My data using exact same command structure:

run_midas.py snps ./test -1 ./ERR866577_1.fastq.gz -2 ./ERR866577_2.fastq.gz -n 100 --species_id 57327

MIDAS: Metagenomic Intra-species Diversity Analysis System
version 1.0.0; github.com/snayfach/MIDAS
Copyright (C) 2015-2016 Stephen Nayfach
Freely distributed under the GNU General Public License (GPLv3)

===========Parameters===========
Script: run_midas.py snps
Output directory: ./test
Remove temporary files: False
Pipeline options:
  -build bowtie2 database of genomes
  -align reads to bowtie2 genome database
  -use samtools to generate pileups and call SNPs
Database options:
  -include specified species id(s): ['57327']
Read alignment options:
  -input reads (1st mate): ./ERR866577_1.fastq.gz
  -input reads (2nd mate): ./ERR866577_2.fastq.gz
  -alignment speed/sensitivity: very-sensitive
  -number of reads to use from input: 100
  -number of threads for database search: 1
SNP calling options:
  -minimum alignment percent identity: 94.0
  -minimum mapping quality score: 20
  -minimum base quality score: 30
  -minimum read quality score: 20
  -trim 0 base-pairs from read-tails

Building database of representative genomes
  total genomes: 1
  total contigs: 107
  total base-pairs: 4029290

Error encountered executing:
apps/MIDAS/bin/Linux/bowtie2-build ./test/snps/temp/genomes.fa ./test/snps/temp/genomes

Error message:
Warning: Empty fasta file: './test/snps/temp/genomes.fa'
Warning: All fasta inputs were empty
Error: Encountered internal Bowtie 2 exception (#1)
Command: bowtie2-build --wrapper basic-0 ./test/snps/temp/genomes.fa ./test/snps/temp/genomes 

However at this moment running apps/MIDAS/bin/Linux/bowtie2-build ./test/snps/temp/genomes.fa ./test/snps/temp/genomes will execute the command no problem.

Re the running two commands at the same time, I wasn't running those at the same time. Simply showing you that both of those commands failed. Didn't matter if it was the automated pipeline vs specifying a species id.

[snayfach]

I wonder if this is related to the species_id. Could you try running the test data on your specified species:
cd MIDAS/test
run_midas.py snps ./sample -1 ./test.fq.gz -n 100 --species_id 57327

This might also be related to file permissions. Could you try running your data, but use this specified directory for your output (replace '/path/to' with the appropriate path):
cd MIDAS/test
run_midas.py snps ./output -1 /path/to/ERR866577_1.fastq.gz -2 /path/to/ERR866577_2.fastq.gz -n 100 --species_id 57327

[snayfach]

I just tried running midas with --species_id 57327 and the test data and it worked fine. So the problem should be either your specified output directory or your specified input files (-1 and -2).

[ddomman]

It seems like it is a permissions issue. When I run the analysis from the test folder everything works. Switching to the analysis folder, it hits that error. I tried changing the permissions for the folder and still didn't work. I wonder why the python call for the script has the error whereas the call from the command line doesn't.

[snayfach]

Very strange. Could you try running it from your home directory where there shouldn't be any permissions issues? Does it work there?

[ddomman]

Sorry, was away for a few days. The pipeline works from my home directory. I'm on a large server, with multiple drives if that helps, but I've never had any problems before with installing software on one and running the analysis on another drive.

[snayfach]

Thanks for checking this. If you end up tracking down what's happening, I'd be happy to build in a check into the software (or a fix).

Stephen