KmerID

AIM: For Species identification. Compares FastQ reads (also fasta files) against custom databases. Resutls come in proper table format without
any confusing names and formats. 

1. Create your own species database using downloaded (or your own) genomes locally and create KmerID database using KmerID.
2. Run KmerID to compare your Fastq reads against the custom database
3. KmerID also comes with custom databases for few species

KmerID will compare your fastq reads against one or more sets of reference
genomes.

There is no Conda receipe for KmerID but installation is very easy in a conda environment. 

Link to GitHub:
https://github.com/phe-bioinformatics/kmerid

Why/When should I use it? 
(From hands-on experience) We had Vibrio Alginolyticus samples. 

1. Method 1: When compared using Kraken2 only 50-60% of the reads mapped to Vibrio alginolyticus. 
And, the rest was unknown but Vibrio genus. 
2. Method 2: The result was similar when used Kaiju database also. 
3. Method 3: BLAST de nova assembly against NCBI NR database was a mess due to high numnbe of query contigs and too many results. 
4. Ref-Mapping the reads using bwa. 
   Pick the unmapped reads
   build contigs
   BLAST against NR database
   Lots of downstream analysis to filter and format the taxa to get any clear results. 
  
Quick Solution: KmerID
I build a local database using selected genomes from NCBI (maximum 15 genomes of a Vibrio alginolyticus). The results showed that 
the sample we have is ~80% Vibrio alginolyticus. 
It takes 10mins to download the genomes for database. 2 mins to build the database and another 1 minute to compare your reads to database.

Conclusion: Kraken2 and Kaiju are not enough.