You could identify the genes in your genomes using PROKKA/PRODIGAL, cluster them with OrthoMCL (since you have fewer genomes) or CD-HIT (much quicker), align the core genes in core clusters (clusters with genes in all your genomes) separately with MUSCLE (or any other alignment tools), concatenate the individual cluster alignments and identify SNP sites in the concatenated alignment using SNP_SITES (https://github.com/sanger-pathogens/snp_sites). After getting the alignment of SNPs, you can use RAxML to build your phylogeny.

There is a tool called ROARY (https://github.com/sanger-pathogens/Roary) that combine all these steps which you might find interesting. Here is the paper that describes the tool: "Roary: Rapid large-scale prokaryote pan genome analysis",

Andrew J. Page et al, Bioinformatics, (2015). doi: http://dx.doi.org/10.1093/bioinformatics/btv421

If you would like some background on SNP calling, a great resource is the GATK paper http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4243306/. If you want a specific pipeline I can recommend https://peerj.com/articles/cs-20/. This is a huge question though and these recommendations are only a very brief introduction.

 @Andrei. By 'extracting' I mean SNP calling. While concatenating I mean substituting   part of the aligned nucleotides with the reference nucleotides. The issues really are how to go about the steps you suggested and easy to use softwares!

@Oscar. Thanks. Please let me ask if the galaxy tools will also accept contigs as input files.? Because,after going through the suggested article I have the impression that may be only reads will be accepted as input file. But I have the contigs of my bacteria.

Hello Kayode,

Can you explain your data set a little more, I think it would be useful to answering your question. Are all of your 14 salmonellae aligned together by contig or do you just have separate files with contigs for each salmonellae separately. Also why do you want to use SNPs? The genome size for salmonellae is small enough (4.5Mbps) that several phylogenetic programs can handle the complete genome for 14 samples. You could do a phylogeny on the whole genomes or just use SNPs, just wondering why you want to use a SNP only data set.

Best Regards

Hi Max

Yes, have separate files with contigs for each salmonella separately. I am considering SNP data based on my understanding of literature. I will appreciate further education on the phylogeny based on whole genomes.

@Oscar. Thanks very much.

Hello Kayode,

The first thing you need to do is align your contigs. My suggestion is to align your contigs to a reference genome since there is a reference genome for salmonella. After you have an alignment you can use various different phylogenetic programs to build a tree using whole sequence data or SNP data. The one issue with SNP only data is that your branch lengths will be over estimated due to acquisition bias. Below is a paper on this topic, but the first thing you need to worry about is alignment and GATK or galaxy will both work for this.

https://www.researchgate.net/publication/280582868_Short_Tree_Long_Tree_Right_Tree_Wrong_Tree_New_Acquisition_Bias_Corrections_for_Inferring_SNP_Phylogenies

Best Regards

Article Short Tree, Long Tree, Right Tree, Wrong Tree: New Acquisiti...

Try using Gegenees Program: http://www.gegenees.org/

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0039107

You could identify the genes in your genomes using PROKKA/PRODIGAL, cluster them with OrthoMCL (since you have fewer genomes) or CD-HIT (much quicker), align the core genes in core clusters (clusters with genes in all your genomes) separately with MUSCLE (or any other alignment tools), concatenate the individual cluster alignments and identify SNP sites in the concatenated alignment using SNP_SITES (https://github.com/sanger-pathogens/snp_sites). After getting the alignment of SNPs, you can use RAxML to build your phylogeny.

There is a tool called ROARY (https://github.com/sanger-pathogens/Roary) that combine all these steps which you might find interesting. Here is the paper that describes the tool: "Roary: Rapid large-scale prokaryote pan genome analysis",

Andrew J. Page et al, Bioinformatics, (2015). doi: http://dx.doi.org/10.1093/bioinformatics/btv421

Thanks all! More suggestions are still welcome.

This may be late, but I will respond anyway ....

There are a lot of interesting answers in here and from the diversity of responses you can tell there are multiple different ways of generating and comparing SNPs among bacterial genomes. Unfortunately, there is yet to be consensus as to a single set of criteria or a clearly dominant approach. Most scientists cobble together a set of tools into a 'pipeline' that suits there hypothesis, works well with the kind of data they have and tools they have experience/expertise on.

Back to your question, constructing a phylogenetic tree for bacterial genomes involves a number of steps:

1) Select the number of genomes, you want to have in your tree.

2) Create an alignment of said genomes

3) Post-processing of your genome alignment to remove poorly aligned bases and/or gaps; (you may need to re-format your alignment depending on the phylogenetic software you will use in step 4)

4) Generate a phylogenetic tree using a software package you choose that applies using an algorithm that works well with your data.

5) View and/or add labels, etc to generated tree using phylogenetic tree viewers

For each of these steps there are multiple tools, some of which are mentioned in this thread that will get the job done. But it is very rare to find a single software that will do all the above steps for you.

The free program kSNP (Gardner, S. N. and B. G. Hall. 2013 When whole-genome alignments just won't work: kSNP v2 software for alignment-free SNP discovery and phylogenetics of hundreds of microbial genomes. PLoS One 8(12): e81760. doi:10.1371/journal.pone.0081760 ) was designed to do exactly what you want to do.  The most recent version is kSNP3 (Gardner, S.N., T. Slezak, and B.G. Hall. 2015. kSNP3.0: SNP detection and phylogenetic analysis of genomes without genome alignment or reference genomes. Bioinformatics 31: 2877-2878 doi: 10.1093/bioinformatics/btv271.)

kSNP identifies SNPs from a set of genome sequences (closed genomes sequences, genome assemblies or even raw reads) and estimates parsimony, ML and NJ trees.  A study of the accuracy of those trees was published recently (Hall, B.G. 2015 Effects of sequence diversity and recombination on the accuracy of phylogenetic trees estimated by kSNP. Cladistics 32: 90-99 (doi: 10.1111/cla.12113).)

kSNP does not require any reference genomes, however if one or more references genomes are included in the data set then kSNP can automatically annotate the SNPs that it finds.  kSNP is available for mac OS X and Linux operating systems.

Hope this helps.

So this is a loaded question!  There are so many different pipelines using different programs.  So still perform denovo but if you have a good reference sequence sequence, I would do reference mapping as well, with programs such as bowtie2, BWA, or SMALT.  GATK also has a "best-practices" guideline for whatever kind of samples you have (i.e., somatic, germline).  Then there are lots of different variant callers, such as the ones in GATK (haplotype caller, or unified genotyper). I usually use Freebayes, but it can depend on the type of samples you're analyzing.  I do SNP calling on whole genomes of bacterial pathogens, which can get tricky since there is no database for calling variants for bacteria.  But do some research into the newest latest tools.  One thing i want to do with my research is come up with a more generalized pipeline for everyone to use and to help made variant calling more reproducible.

Hi Kayode,

Have you heard about the http://www.genomicepidemiology.org/ the centre for genomic epidemiology? The aim of this centre domiciled in DTU my host institution for AMR fellowship by FlemingFund was established to provide the scientific foundation for future internet-based solutions where a central database will enable simplification of total genome sequence information and comparison to all other sequenced including spatial-temporal analysis.

You can achieve this using the CSI phylogeny tool https://cge.cbs.dtu.dk/services/CSIPhylogeny/ ( SI Phylogeny calls SNPs, filters the SNPs, does site validation and infers a phylogeny based on the concatenated alignment of the high quality* SNPs. ). Then the https://cge.cbs.dtu.dk/services/NDtree/ NDtree constructs phylogenetic trees from Single-End or Pair-End FASTQ files

There is a course on Coursera by DTU https://www.coursera.org/learn/wgs-bacteria/

Please feel free to reach if you want more clarification hope this is useful.

I use Parsnp.