KAIJUXREPORT

Author: Dimitar Kenanov – kenietz78 at gmail dot com

kaijuxReport.pl - program which creates a report file for kaijux. Its input is kaijux output file and hash file with headers of all sequences in the KaijuxDB.

Kaijux which is a part of the Kaiju metagenomics project which classifies reads or contigs: https://github.com/bioinformatics-centre/kaiju

The report file follows the Kaiju scheme but with 5 columns:

% - percent of all reads reads - number of reads. Or contigs in the case of classifying contigs id - protein sequence ID desc - protein sequence description species – species name

At the end of the report can be found those:

DET:S: - Detected seqs TOT:R: - Total Num Read/Contigs TOT:C: - Total Num Classifiled Reads/Contigs TOT:U: - Total Num Unclassified Reads/Contigs

A. Using kaijuxReport.pl

kaijuxReport.pl -h HEADERS_HASH_FILE -k KAIJUX_OUTPUT > KAJ_REPORT_FILE

Instructions on how to build HEADERS_HASH_FILE and KAIJUX DB are provided below. One can create a full DB with all the hypothetical proteins or one might decide to exlude them.

Example of creating of KAIJUX DB with Bacteria from refseq: ftp://ftp.ncbi.nlm.nih.gov/refseq/release/bacteria/

B. Create a regular KAIJUX DB

1. Get all proteins from REFSEQ, e.g all 'bacteria.nonredundant_protein*.faa.gz' files.

2. Cat all of them into a single file e.g 'all.fa'

3. Get the headers with:

'grep -P '^>' all.fa > headers.txt'

4. Create a KAIJUX DB with the appropriate tools provided by KAIJU

mkbwt -n 5 -a ACDEFGHIKLMNPQRSTVWY -o KJX_DB all.fa

mkfmi KJX_DB

5. Create a HEADERS_HASH_FILE with 'create_HEADERS_HASH_FILE.pl'

create_HEADERS_HASH_FILE.pl -h headers.txt -o HEADERS_HASH_FILE

C. Create KAIJUX DB without Hypothetical proteins

1. Get all proteins from REFSEQ, e.g all 'bacteria.nonredundant_protein*.faa.gz' files

2. Cat all of them into a single file e.g 'all.fa'

3. Get the headers with:

grep -P '^>' all.fa > headers.txt

4. Filter out the 'hypotetical protein' headers with:

grep -v 'hypothetical protein' headers.txt > filtered_headers.txt

5. Filter out the 'hypotetical protein' sequences with filter_fasta_by_name.pl

filter_fasta_by_name.pl -h filtered_headers.txt -fi all.fa -fo filtered.fa

6. Create a KAIJUX DB with the appropriate tools provided by KAIJU

mkbwt -n 5 -a ACDEFGHIKLMNPQRSTVWY -o KJX_DB filtered.fa

mkfmi KJX_DB

7. Create a HEADERS_HASH_FILE with 'create_HEADERS_HASH_FILE.pl'

create_HEADERS_HASH_FILE.pl -h filtered_headers.txt -o HEADERS_HASH_FILE