bowtie error

[Nir Werbner]

hi,

i installed humann2 via pip install, and metaphlan2 later in a conda environment.

there were missing databases, which i finally found in the metaphlan2 group and downloaded them.

so now it works, until it gets stuck here:

Running bowtie2-build ........

CRITICAL ERROR: Can not find file humann2/sample20_humann2_temp/sample20_bowtie2_index.rev.1.bt2

any ideas?

thanks,

nir

[Lauren McIver]

Hello - I am not sure exactly what is the cause of the bowtie2 error you are seeing. Could you please provide your humann2 log file? Feel free to email it to me directly if you do not want to post it to the list. I would be happy to help debug.

Thanks!

Lauren

[Nir Werbner]

hi,

i wonder if it could be the bowtie database that i downloaded from somewhere in this group.

from some reason the conda installation of metaphlan2 didn't install the databases at all. was it supposed to?
attached the log file.

thanks

[Lauren McIver]

Hi - It looks like you are all set with the databases. It seems that bowtie2 is running into an error during the build step. I don't see any error from bowtie2 in the log file. Could you try running the last command (bowtie2-build) in the log file directly on the command line? This might give us more information as to what the error might be.

Thanks!

Lauren

[Nir Werbner]

(humann2) poranmi@ahalama2:~/metatranscriptomics_led/humann2/sample19_humann2_temp$ /home/korenlab/shared/miniconda3/envs/humann2/bin/bowtie2-build -f /home/avnilab/poranmi/metatranscriptomics_led/humann2/sample19_humann2_temp/sample19_custom_chocophlan_database.ffn /home/avnilab/poranmi/metatranscriptomics_led/humann2/sample19_humann2_temp/sample19_bowtie2_index

Settings:

  Output files: "/home/avnilab/poranmi/metatranscriptomics_led/humann2/sample19_humann2_temp/sample19_bowtie2_index.*.bt2"

  Line rate: 6 (line is 64 bytes)

  Lines per side: 1 (side is 64 bytes)

  Offset rate: 4 (one in 16)

  FTable chars: 10

  Strings: unpacked

  Max bucket size: default

  Max bucket size, sqrt multiplier: default

  Max bucket size, len divisor: 4

  Difference-cover sample period: 1024

  Endianness: little

  Actual local endianness: little

  Sanity checking: disabled

  Assertions: disabled

  Random seed: 0

  Sizeofs: void*:8, int:4, long:8, size_t:8

Input files DNA, FASTA:

  /home/avnilab/poranmi/metatranscriptomics_led/humann2/sample19_humann2_temp/sample19_custom_chocophlan_database.ffn

Building a SMALL index

Reading reference sizes

  Time reading reference sizes: 00:00:00

Calculating joined length

Writing header

Reserving space for joined string

Joining reference sequences

  Time to join reference sequences: 00:00:00

bmax according to bmaxDivN setting: 1730681

Using parameters --bmax 1298011 --dcv 1024

  Doing ahead-of-time memory usage test

  Passed!  Constructing with these parameters: --bmax 1298011 --dcv 1024

Constructing suffix-array element generator

this is the terminal output.

it still creates only 4  files, without the rev files, although rev index files do appear in the db_v20 folder.
 

[Lauren McIver]

Hello - Thank you for posting the additional output. It looks like bowtie2 is ending before the build is complete. What version of bowtie2 are you running? I tried running your build with bowtie2 v2.2.1 and it seemed to work okay. Please let me know your version and I will test with it.

Thanks!

Lauren

[Nir Werbner]

bowtie2-align-s version 2.3.4.1

[Lauren McIver]

Hello - Thanks for providing the version you are running. I tried running the same build command with the same files and the same version of bowtie2 and it ran without error (and created the expected six output files). It looks like your bowtie2 build run stops in the middle of the build but does not generate an error. Can you double check your bowtie2 install is okay? Maybe try uninstalling and reinstalling bowtie2?

Thanks!

Lauren

[Nir Werbner]

hi,

could it be that the databases sre not suitable for my bowtie2 version?

i tried reinstalling everything through conda, and when it built the databases, again it did not build the .rev databases.

maybe it will be better if i installed it not through conda?

[Lauren McIver]

Hi - Yes, you could try reinstalling the humann2 nucleotide database and see if this resolves the issue. A quick check would be to look at the input file to the bowtie2-build command to make sure it is of the correct fasta format. Other than that, maybe installing bowtie2 with a different method (instead of conda) like you had mentioned would help. It is a bit difficult to know what might be going on since bowtie2 ends early without an error and I can't seem to replicate the behavior with that version of bowtie2. Please try these two things and let me know if you continue to have issues.

Thanks!

Lauren