how to run pathway analysis for KEGG

[yun li]

Hi,

I have run humann2 on the default databases. However, now I want to check the pathway abundance of KEGG. How can I do this? I've asked this question before, 

Here is the answer I got from Eric 

"Yes, this is possible. If you use the "regroup_table" script to convert UniRef50 gene family abundance to approximate KO abundance, you can then follow the legacy database instructions for making a KEGG module or KEGG pathway table starting from gene-level input:

https://bitbucket.org/biobakery/humann2/src/tip/doc/UserManual.md?fileviewer=file-view-default#markdown-header-legacy-databases

Thanks,

Eric

"

But I still don't know the exact steps how to do this.

When I tried to run regroup_table, it says invalid choice: 

$ humann2_regroup_table --input log2GeneFamilyQuantileNormalized.tsv --groups uniref50_ko --output uniref50_ko_known.tsv

usage: humann2_regroup_table [-h] [-i INPUT] [-g {uniref50_ec,uniref50_rxn}]

                             [-c CUSTOM] [-r] [-f {sum,mean}] [-u {Y,N}]

                             [-p {Y,N}] [-o OUTPUT]

humann2_regroup_table: error: argument -g/--groups: invalid choice: 'uniref50_ko' (choose from 'uniref50_ec', 'uniref50_rxn')

When I run the following command, it also report error.

$humann2 --input geneFamilies.tsv --output KEGGRes --id-mapping legacy_kegg_idmapping.tsv --pathways-database humann-0.99/data/keggc

Output files will be written to: /home/myfolder/Project1/12.gene2regroup/KEGGRes

Process the sam mapping results ...

Traceback (most recent call last):

  File "//home/myfolder/.local/bin/humann2", line 9, in <module>

    load_entry_point('humann2==0.6.2', 'console_scripts', 'humann2')()

  File "/home/myfolder/.local/lib/python2.7/site-packages/humann2-0.6.2-py2.7.egg/humann2/humann2.py", line 906, in main

    args.input, alignments, unaligned_reads_store, keep_sam=True)

  File "/home/myfolder/.local/lib/python2.7/site-packages/humann2-0.6.2-py2.7.egg/humann2/search/nucleotide.py", line 251, in unaligned_reads

    if int(info[config.sam_flag_index]) & config.sam_unmapped_flag != 0:

ValueError: invalid literal for int() with base 10: 'A468xx000-000'

Thanks,

Yun

[Eric Franzosa]

Hi Yun,

We've started only bundling HUMAnN2 with a couple of default mapping files, as the full set was getting rather large. You can download the UniRef50-KEGG mapping here:

https://bitbucket.org/biobakery/humann2/src/tip/humann2/data/misc/map_infogo1000_uniref50.txt.gz

You will pass this mapping to regroup_table as a custom mapping using the "-c" flag.

====

After doing this, you'll have KO abundance, which actually makes the legacy KEGG pathway quantification fairly simple:

humann2 --input KO.tsv --output $OUTPUT_DIR --pathways-database humann1/data/keggc

The full legacy instructions assume you want to completely replicate a HUMAnN1 run (i.e. starting from raw reads). You're able to skip some of this since you already have UniRef50 abundance (and can regroup to get KOs).

Thanks,

Eric

[yun li]

Hi Eric,

should I use map_ko_uniref90.txt.gz in this case, and not use -g flag? and  "KO.tsv" in the second command is the output from humann2_regroup_table?  

[yun li]

Hi Eric,

I runned humann2_regroup_table  using  map_ko_uniref50.txt. Is this right?

Then I run KEGG pathway analysis, However, I still get error from the second step.

$humann2_regroup_table --input log2GeneFamilyQuantileNormalized.tsv -c  map_ko_uniref50.txt --output uniref50_ko_known.tsv

$humann2 --input uniref50_ko_known.tsv --output KEGGRes --pathways-database humann-0.99/data/keggc

Process the sam mapping results ...

Traceback (most recent call last):

  File "//home/myfolder/.local/bin/humann2", line 9, in <module>

    load_entry_point('humann2==0.6.2', 'console_scripts', 'humann2')()

  File "/home/myfolder/.local/lib/python2.7/site-packages/humann2-0.6.2-py2.7.egg/humann2/humann2.py", line 906, in main

    args.input, alignments, unaligned_reads_store, keep_sam=True)

  File "/home/myfolder/.local/lib/python2.7/site-packages/humann2-0.6.2-py2.7.egg/humann2/search/nucleotide.py", line 251, in unaligned_reads

    if int(info[config.sam_flag_index]) & config.sam_unmapped_flag != 0:

ValueError: invalid literal for int() with base 10:' A468xx000-000''

[yun li]

Hi, I feel really confused about how to rerun the KEGG abudance. When I added flag "--input-format genetable", the second command "humann2 --input uniref50_ko_known.tsv --output KEGGRes --pathways-database humann-0.99/data/keggc" worked. however, the output is really confusing. my input "uniref50_ko_known.tsv" is a tab separated file with each run a ko number and each column a sample, this file is generated from humann2_regroup_table. But the output pathabundance have only one column, any idea what happened?

Thanks,

Yun

[Eric Franzosa]

Hi Yun,

HUMAnN2 proper only thinks about processing one sample at a time, so the fact that you're providing a multi-sample TSV explains both 1) why the format wasn't immediately detected and 2) why the output contains fewer columns that you expected.

If you want to reprocess a merged table through HUMAnN2, you need to split up the columns, process them separately, and then recombine them. You can see this procedure spelled out in the context of a PICRUSt application here:

https://bitbucket.org/biobakery/humann2/wiki/Home#markdown-header-picrust-output

Thanks,

Eric

[yun li]

I was able to run the program if I provide one sample at a time. However, what actually is the pathabundance value from the generated file. 

for example: ko05340 has  K10887 K10603 K10989 K10988.... proteins. The pathabundance value for ko05340 is not the sum nor the mean of the proteins. How ko05340 value was calculated?

Thanks,

Yun

[Eric Franzosa]

Please see the subsection "Pathway/module coverage and abundance" from the HUMAnN1 paper:

http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1002358

These methods have remained largely unchanged in HUMAnN2.

Thanks,

Eric