llama runs on MacOS and Linux. The conda environment recipe may not build on Windows and is not supported but can be run using the Windows subsystem for Linux.

  1. Some version of conda, we use Miniconda3. Can be downloaded from here
  2. Your input query file with a row for each sequence name you want to analyse/ create local trees for. These can be present in the big tree already or in the fasta file you supply
  3. Optional fasta file if there are sequences you want to add into the tree
  4. A directory of data containing the following files:
    • global.tree: a large tree that you want to place your sequences in
    • alignment.fasta: an alignment file with the fasta sequences used to make the tree
    • metadata.csv: associated metadata with minimally the name of the sequences in the tree/ alignment and a lineage designation The names of the tips of the tree, the sequence ids in the alignment and the column you select as --data-column in the metadata must match

Install llama

  1. Clone this repository and cd llama
  2. conda env create -f environment.yml
  3. conda activate llama
  4. python setup.py install

Note: we recommend using llama in the conda environment specified in the environment.yml file as per the instructions above. If you can’t use conda for some reason, dependency details can be found in the environment.yml file.

Check the install worked

Type (in the llama environment):


and you should see the help menu of llama printed.

Updating llama

Note: Even if you have previously installed llama, as it is being worked on intensively, we recommend you check for updates before running.

To update:

  1. conda activate llama
  2. git pull
    pulls the latest changes from github
  3. python setup.py install
    re-installs llama
  4. conda env update -f environment.yml
    updates the conda environment

  5. Activate the environment conda activate llama
  6. Run llama

Example usage:

llama -i <input.csv> -f <input.fasta> -d <path/to/data> [options]

Full usage:

usage: llama -i <input.csv> -d <path/to/data> [options]

llama: Local Lineage And Monophyly Assessment

optional arguments:
  -h, --help            show this help message and exit
  -i QUERY, --input QUERY
                        Input csv file with minimally `name` as a column
                        header. Alternatively, `--input-column` can specifiy a
                        column name other than `name`
                        Generate a query from the metadata file supplied.
                        Define a search that will be used to pull out
                        sequences of interest from the large phylogeny. E.g.
                        -fm country=Ireland sample_date=2020-03-01:2020-04-01
  -f FASTA, --fasta FASTA
                        Optional fasta query. Fasta sequence names must
                        exactly match those in your input query.
  -a, --align-sequences
                        Just align sequences.
  -s SEQS, --seqs SEQS  Sequence file containing sequences used to create the
                        tree. For use in combination with the `--align-
                        sequences` option.
  -ns, --no-seqs        Alignment not available. Note, to work, all queries
                        must already be in global tree.
  -r, --report          Generate markdown report of input queries and their
                        local trees
  --colour-fields COLOUR_FIELDS
                        Comma separated string of fields to colour by in the
  --label-fields LABEL_FIELDS
                        Comma separated string of fields to add to tree report
  --id-string           Indicates the input is a comma-separated id string
                        with one or more query ids. Example:
  -o OUTDIR, --outdir OUTDIR
                        Output directory. Default: current working directory
  -d DATADIR, --datadir DATADIR
                        Local directory that contains the data files
  --tempdir TEMPDIR     Specify where you want the temp stuff to go. Default:
  --no-temp             Output all intermediate files, for dev purposes.
  --input-column INPUT_COLUMN
                        Column in input csv file to match with database.
                        Default: name
  --data-column DATA_COLUMN
                        Column in database to match with input csv file.
                        Default: sequence_name
  --distance DISTANCE   Extraction from large tree radius. Default: 2
  --collapse-threshold THRESHOLD
                        Minimum number of nodes to collapse on. Default: 1
                        Include a selection of representative sequences from
                        lineages present in the local tree. Default: False
  --number-of-representatives NUMBER_OF_REPRESENTATIVES
                        How many representative sequeneces per lineage to keep
                        in the collapsed tree. Default: 5
  --max-ambig MAXAMBIG  Maximum proportion of Ns allowed to attempt analysis.
                        Default: 0.5
  --min-length MINLEN   Minimum query length allowed to attempt analysis.
                        Default: 10000
  -n, --dry-run         Go through the motions but don't actually run
  -t THREADS, --threads THREADS
                        Number of threads
  --verbose             Print lots of stuff to screen
  --outgroup OUTGROUP   Optional outgroup sequence to root local subtrees.
                        Default an anonymised sequence that is at the base of
                        the global SARS-CoV-2 phylogeny.
  -v, --version         show program's version number and exit


Curate the input sequences into an alignment padded against an early lineage A reference:

llama -a -s your_input_sequences.fasta

Generate a report with your sequences summarised:

llama -r -i test.csv -f test.fasta -d <path/to/data>

Generate a report with a custom set of sequences defined by the metadata file supplied. After the -fm or --from-metadata argument, one or more columns in the metadata and search pattern to match can be described. A special case exists if a date range is detected (colon separated dates). The required date format is YYYY-MM-DD. The format of this search is as follows:

--from-metadata column1=value1 column2=YYYY-MM-DD:YYYY-MM-DD

For example, the following command will pull out sequences from Ireland with samples between 2020-03-01 and 2020-04-1, provided that information exists metadata.csv file found in the data directory.

llama -r -fm country=Ireland sample_date=2020-03-01:2020-04-01 -d <path/to/data>

Include a selection of representative sequences for each lineage present in the local tree:

llama -i test.csv --fasta test.fasta --datadir <path/to/data> \
--lineage-representatives \
--number-of-representatives 5

Analysis pipeline


  • From the input csv (<query>), llama attempts to match the ids with ids in the metadata.csv.

  • If the id matches with a record, the corresponding metadata is pulled out.

  • If the id doesn’t match with a record and a fasta sequence with that query id has been provided, it’s passed into a workflow (find_closest_in_db.smk) to identify the closest sequence. In brief, this search consists of quality control steps that maps the sequence against a reference (an early, anonymised sequence from lineage A at the root of the global tree), pads any indels relative to the reference and masks non-coding regions. llama then runs a minimap2 search against the alignment.fasta file and finds the best hit to the query sequence.

  • The metadata for the closest sequences are then also pulled out of the large metadata.csv.

  • Combining the metadata from the records of the closest hit and the exact matching records found in the csv, llama queries the large global.tree phylogeny. The local trees around the relevant tips are pruned out of the large phylogeny, merging overlapping local phylogenys as needed. By default, llama pulls out a local tree two above the query id tips, but this can be customised with the --distance argument if larger or smaller trees are desired.

  • If these local trees contain “closest-matching” tips that have been found based on the input fasta file, the sequence records for the tips on the tree and the sequences of the relevant queries are added into an alignment. llama then checks what lineages are present in the local tree and flags a maximum of 10 sequences per lineage to retain the surrounding context of the tree. Any peripheral sequences coming off of a polytomy that are not flagged and are not the query sequences are collapsed to a single node and summaries of the tip’s contents are output. An outgroup sequence from the base of the tree at lineage A is added into the alignment.

  • After collapsing the nodes, llama runs iqtree on the new alignment, with the outgroup and query sequences in, and then prunes off the outgroup sequence.

  • llama then annotates this new phylogeny with lineage assignments and can produce a report (COMING SOON!).


  • Catchment trees around the query sequences (uncollapsed)
  • Collapsed local trees (containing query sequences if optional fasta file supplied) with a representative set of sequences from surrounding lineages and query tips uncollapsed


llama makes use of datafunk and clusterfunk functions which have been written by members of the Rambaut Lab, specificially Rachel Colquhoun, JT McCrone, Ben Jackson and Shawn Yu.

llama runs a java implementation jclusterfunk written by Andrew Rambaut.

baltic by Gytis Dudas is used to visualize the trees.



Heng Li, Minimap2: pairwise alignment for nucleotide sequences, Bioinformatics, Volume 34, Issue 18, 15 September 2018, Pages 3094–3100, https://doi.org/10.1093/bioinformatics/bty191


L.-T. Nguyen, H.A. Schmidt, A. von Haeseler, B.Q. Minh (2015) IQ-TREE: A fast and effective stochastic algorithm for estimating maximum likelihood phylogenies.. Mol. Biol. Evol., 32:268-274. https://doi.org/10.1093/molbev/msu300

D.T. Hoang, O. Chernomor, A. von Haeseler, B.Q. Minh, L.S. Vinh (2018) UFBoot2: Improving the ultrafast bootstrap approximation. Mol. Biol. Evol., 35:518–522. https://doi.org/10.1093/molbev/msx281

Stéphane Guindon, Jean-François Dufayard, Vincent Lefort, Maria Anisimova, Wim Hordijk, Olivier Gascuel, New Algorithms and Methods to Estimate Maximum-Likelihood Phylogenies: Assessing the Performance of PhyML 3.0, Systematic Biology, Volume 59, Issue 3, May 2010, Pages 307–321, https://doi.org/10.1093/sysbio/syq010


Köster, Johannes and Rahmann, Sven. “Snakemake - A scalable bioinformatics workflow engine”. Bioinformatics 2012.

Software versions

  • python=3.6
  • snakemake-minimal=5.13
  • iqtree=1.6.12
  • minimap2=2.17-r941
  • pandas==1.0.1
  • pytools=2020.1
  • dendropy=4.4.0