Humann2 stucks at bowtie2 alignment step

[Changan Liu]

Hello!

I just applied Humann2 to analyze my filtered metagenomic sequencing data (.fq files) for the first time.

For my samples (the .fq files with around 20 GB), it takes about one day to finish running one sample on our server (with threads 24) usually.

However, for some samples, it takes too long (more than 3 days). And I can not get results until now. I checked the log files, it seems that it stucks at the bowtie2 alignment step. The bottom lines of the log file are listed below:

10/10/2019 11:55:21 AM - humann2.humann2 - INFO: TIMESTAMP: Completed     custom database creation     :     2     seconds
10/10/2019 11:55:21 AM - humann2.search.nucleotide - INFO: Running bowtie2-build ........
10/10/2019 11:55:21 AM - humann2.utilities - DEBUG: Using software: /storage1/changanliu/software/conda/miniconda3/envs/humann2/bin/bowtie2-build
10/10/2019 11:55:21 AM - humann2.utilities - INFO: Execute command: /storage1/changanliu/software/conda/miniconda3/envs/humann2/bin/bowtie2-build -f /storage1/changanliu/work/For_BXW/Humann2/AOM/AOM673_humann2_temp_IWS4Er/AOM673_custom_chocophlan_database.ffn /storage1/changanliu/work/For_BXW/Humann2/AOM/AOM673_humann2_temp_IWS4Er/AOM673_bowtie2_index
10/10/2019 11:56:11 AM - humann2.humann2 - INFO: TIMESTAMP: Completed     database index     :     50     seconds
10/10/2019 11:56:11 AM - humann2.search.nucleotide - DEBUG: Nucleotide input file is of type: fastq
10/10/2019 11:56:11 AM - humann2.utilities - DEBUG: Using software: /storage1/changanliu/software/conda/miniconda3/envs/humann2/bin/bowtie2
10/10/2019 11:56:11 AM - humann2.utilities - INFO: Execute command: /storage1/changanliu/software/conda/miniconda3/envs/humann2/bin/bowtie2 -q -x /storage1/changanliu/work/For_BXW/Humann2/AOM/AOM673_humann2_temp_IWS4Er/AOM673_bowtie2_index -U /storage1/SHARED_DATA/Metagenome/BXW_smoking201904/metagenome_201906/02.Host_Alignment/S673_H7FYVDSXX_L3/S673_H7FYVDSXX_L3.umhost.fq -S /storage1/changanliu/work/For_BXW/Humann2/AOM/AOM673_humann2_temp_IWS4Er/AOM673_bowtie2_aligned.sam -p 24 --very-sensitive

The screenshot of the htop command on our server running Humann2 is as following:

Is this situation normal or something is wrong? Any ideas?

Thank you very much for your help!

Best regards

[Eric Franzosa]

Hi Changan,

Are you taking steps to QC your reads before running HUMAnN2? Sometimes hanging during alignment can be caused by extensive low-complexity and/or ambiguous region in the queries. The bowtie2 mapping step tends to be very consistent in our evaluations (20-30K reads/sec on 8 cores), so it does sound like something might be different about this sample.

Thanks,

Eric

--
You received this message because you are subscribed to the Google Groups "HUMAnN Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to humann-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/humann-users/03870d52-b378-4a23-b00f-3f36014a8a04%40googlegroups.com.

[Changan Liu]

Dear Eric :

Thank you very much for your reply.

I have taken steps to QC my raw reads. And I used Kraken2 for my reads after QC to get OTU table before applying HUMAnN2. Now I have tried 5 samples. 3 samples are fine but the rest 2 samples are hanging on the bowtie2 alignment step. I will check the samples.

Thank you!

Best regards,

Changan Liu

[Eric Franzosa]

Can you clarify how you're plugging Kraken2 into HUMAnN2?

And also share the command you're executing that's leading to the hanging run?

Thanks,

Eric

--
You received this message because you are subscribed to the Google Groups "HUMAnN Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to humann-users...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/humann-users/684fdeb2-ab59-40c0-ae91-e44dbbc6c1bc%40googlegroups.com.

[Changan Liu]

I didn't plug Kraken2 into HUMAnN2. The two processes are independent for me. For getting OTU tables, I used Kraken2. For pathways and functions, I am applying HUMAnN2.

The command I executed :

nohup humann2 --threads 24 --remove-temp-output --input /storage1/SHARED_DATA/Metagenome/BXW_smoking201904/metagenome_201906/02.Host_Alignment/S673_H7FYVDSXX_L3/S673_H7FYVDSXX_L3.umhost.fq --output-basename AOM673 --output /storage1/changanliu/work/For_BXW/Humann2/AOM > nohup1.out &

Thank you very much for your help!