Skip to content

Latest commit

 

History

History
39 lines (29 loc) · 2.95 KB

FAQ.md

File metadata and controls

39 lines (29 loc) · 2.95 KB
Raw

Reference genomes

Can I have multiple reference genomes?

No. However, you can add multiple assemblies to the asm directory. Assemblies don't have the same error profile as reads and so you might expect some skew in the ultimate phylogeny.

How do I choose a reference genome?

The best reference genome for an outbreak is something in-clade. You might even need to assemble the genome from your own reads before starting. An outgroup is not as related to your clade by definition and so it is probably not the best genome to use. The next best quality is a closed genome, or a genome with a high N50.

How to I include my reference genome in my analysis?

You might have noticed that your reference genome is not included in the final analysis. If you want to include the reference genome in your final tree or SNP matrix, there are about two ways. Either:

  • Copy the assembly into the asm subfolder. There is a more automated way to do that with the set_manage.pl script, using --add-assembly as shown here
  • Find the original raw reads, e.g. Illumina fastq files, and place them into he reads folder. You can directly copy the interleaved files into that folder, or you can run set_manage.pl --add-reads on those interleaved files.

NOTE Keep the assembly in the reference/ subfolder and continue to point -ref to the assembly in the reference subfolder.

High-quality-ness

What are the different ways that SNPs in Lyve-SET have high confidence?

High quality for SNPs indicates that the resulting phylogeny will be high-fidelity. Although some SNPs are discarded that we are less sure about, the SNPs that we are most sure about are retained, and the resulting phylogeny is the best inference.

User Lori Gladney created this flowchart to help understand these points.

  • Detection of troublesome regions such that they are not considered in hqSNP analysis. Currently in v1.0, only phage genes are detected; however other databases could be added in the future, and also I am open to other suggestions. Users can also specify a BED-formatted file to describe regions to mask.
  • Only unambiguous mapping allowed
  • Default 75% consensus and 10x coverage thresholds. These options can be changed when you launch Lyve-SET.
  • Mechanisms to remove clustered SNPs -- not on by default however.
  • Maximum likelihood phylogeny reconstruction. Ascertainment bias is also considered through RAxML v8.
  • Both forward and reverse reads must support each SNP.
  • Each read must have 95% identity to the reference genome. In other words, if the read is 100bp, then only 5 differences in the read can be tolerated before the read is discarded.