Take in a FASTA files => output a codon-based alignment as a BAM file.

Unless I am missing something, we can't run all vs all comparisons between two datasets.

We will need to integrate all tools associated with FLEA into Galaxy. The repository is located here. The pipeline utilizes a workflow language called nextflow. Documentation should already be thorough, but utilizing nextflow's tracing and visualization may aid in determining each step in the process.

https://github.com/veg/hivclustering

(look at the develop branch)

Input (at least CSV file from TN93) => JSON file with the inferred network + some states.

https://github.com/veg/tn93/blob/master/CMakeLists.txt#L45

https://github.com/veg/tn93/blob/master/CMakeLists.txt#L67

This file format was specifically used for 454 sequencing. While 454 sequencing is now dated, we may need it as part of the Galaxy workflow. An alternative is to skip this step entirely and just upload the FASTQ files.

Please check that all the options for hivnetworkcsv have the correct data type. For example, the -s option (sequence alignment) is notated as an hmmer3 file. It should instead be set up to accept FASTA files.

https://github.com/veg/idepi

https://github.com/veg/tn93/blob/master/CMakeLists.txt#L37

Needs to be completed before April 22nd, 2018.

Before wrapping mcmc.jl in issue #8, we should place mcmc.jl in its own repository for use with Galaxy.

Once the repository has been created, a new issue will need to be added to the repo :

Use the BAM format as opposed to FASTA format for input and output.

Identify new dataset to use.

This will require identifying how data munging will be migrated to Galaxy.

https://github.com/stevenweaver/HIV-NGS/blob/master/python/result_processor.py

According to the Swanstrom Lab

Primer IDs that appeared once came from sequence reads that on average had significantly lower quality scores than the scores for the higher frequency reads. This observation suggests that low-abundance Primer ID reads are at least in part the result of misreading of more-abundant Primer IDs that are being resampled (after the PCR amplification step) in the sequencing.

Ruby scripts for the simulation model, consensus cutoff, and consensus creation are provided as supplementary materials in the paper.

We will need to implement the error model in a new script for use with the rest of our pipeline.

https://github.com/veg/BioExt/blob/master/scripts/seqmerge

In order to achieve greater flexibility in the future, investigate whether translating pipeline to CWL is an option.

Once the pipeline is wrapped, the following dashboard (or something like it) should be accessible within Galaxy:

Wrap some subset of aBSREL, BUSTED, RELAX, SLAC, FEL, MEME, FUBAR

• aBSREL
• BUSTED
• FADE
• FEL
• FUBAR
• GARD (still known stochastic errors but the issue is with HyPhy and not the galaxy tool)
• MEME
• RELAX
• SLAC

When setting 'advanced' options in bealign to discard reads that don't match, an error is reliably generated.

From #43

https://github.com/stevenweaver/HIV-NGS/blob/master/hyphy/maketree.bf

The output of bealign is notated as FASTA, but in the history you see output that is (I think) stderr.

FASTA alignment => CSV file of distances gated on threshold.

https://github.com/veg/TN93

https://github.com/veg/HIV-NGS/blob/master/julia/mcmc.jl

https://github.com/veg/hivtrace-viz

https://github.com/stevenweaver/HIV-NGS/blob/master/hyphy/intra_host.bf

TN93 output is not properly displayed as CSV in the viewer (columns not recognized).

Create a visualization plugin of phlotree.js for galaxy. The hivtrace viz could serve as a useful starting point. You can also see me for the mmvc visualization, but it is not PR ready until we follow up with Sam on the supported way to pass multiple datasets to a viz.

Ensure that qfilt has been merged into tools-iuc.

https://github.com/veg/BioExt/blob/master/scripts/translate

In order to wrap individual batch file scripts as tools, we will need to create a conda package out of HyPhy first. @davebx has graciously agreed to taking on this work.

• Two main data generation strategies: (1) amplicon sequencing and (2) shotgun sequencing
• At this time analysis one can choose calling variants or performing within-host evolution analyses including selection detection, rate of evolution, various types of temporal and spatial dynamic analyses. These begin with the following general themes:
• Shotgun data can be analyzed either "read-by-read" using individual reads (i.e., Neher's sliding window approaches) or by reconstructing individual haplotypes.
• There is a number of haplotype reconstruction approaches (nice summary of various approaches is in the introduction section of SAVAGE paper ). Two practical (based on perceived software availability) approaches are:
  1. SAVAGE - overlap graph-based approach. There is a conda recipe for Linux. However it is unusable due to boost versioning. Need to check scalability on large number of reads.
  2. ShoRAH - uses read alignments to a reference genome to reconstruct local haplotypes and then use these to reconstruct global (genome-wide) haplotypes. No conda package. Minimal documentation.
• Once haplotypes are generates multiple alignments are generated with something like MAFFT.
• Tree reconstruction can then be performed with tools such as RAxML or IQTree.
• The subsequent analyses are based on multiple datamonkey tools (such as RELAX) and HyPhy.

Separately we want to implement tools for metagenomic identification of viral sequences along the lines of this paper.