This Jupyter notebook creates a table of COVID-19 virus lineage aliases for existing pangolin lineages.
The script uses a modified pango_aliasor package, which provides tools for aliasing and compressing PANGO lineages of the SARS-CoV-2 virus.
It effectively combines the following files in a manner that allows regularization of the labelling according to user preferences:
https://raw.githubusercontent.com/cov-lineages/pango-designation/master/pango_designation/alias_key.json
https://raw.githubusercontent.com/cov-lineages/pango-designation/master/lineage_notes.txt
A DataFrame is created with the compressed lineage names. The base variants are added as a new column 'alias' to the DataFrame. If a lineage does not have a base variant, the lineage name is used as the alias.
To control this the alias_list is a list of PANGO lineage names that an analyst wants to map to their subvariants. The aliasing is done in such a way as to partition the rows to the lowest common ancestor in the
pangolin hierarchy, i.e. it chooses the most specific alias label that applies to the row.
The user supplied pango_to_voc dictionary maps some PANGO lineage names to their corresponding Variants of Concern (VOC) names.
Finally, rows with lineage names starting with '*' are dropped from the DataFrame, as these lineages are withdrawn. The DataFrame is then saved to a CSV file.
The make_variant_base_map function is defined to create a dictionary that maps each variant to its base variant. The function takes a list of base variants and a boolean flag indicating whether recombinant variants should be included. It creates an instance of the Aliasor class, enables expansion of the aliasor, and then partitions the base variants into their subvariants. If a base variant does not have any subvariants, it is mapped to itself.
The replace_map is created by calling the make_variant_base_map function with the alias_list as the argument. This map is used to replace the lineage names in the table with their base variants.
The compressed_names list is created by compressing the names of the internal structure in the pango_aliasor. This is a very circuitous route to get the names that are simply in lineage_notes first column.
The pango_notes dictionary is populated with notes for each lineage. These notes are also feteched from lineage_notes.