"No markers for contigs in table. Unable to assess binning quality"

[chasemc]

Sometimes the following Autometa entryppoint

autometa-binning \
--kmers ~/kmers.tsv \
--coverages ~/coverages.tsv \
--gc-content ~/gc_content.tsv \
--markers ~/bacteria.markers.tsv \
--output-binning ~/output-binning.tsv \
--output-main ~/output-main.tsv \
--clustering-method hdbscan \
--completeness 20.0 \
--purity 95.0 \
--cov-stddev-limit 25.0 \
--gc-stddev-limit 5.0 \
--taxonomy ~/taxonomy.tsv \
--starting-rank superkingdom \
--rank-filter superkingdom \
--rank-name-filter bacteria

will error out with the warning

No markers for contigs in table. Unable to assess binning quality

This can be frustrating to debug especially when all the input tsv files have aren't empty. The cause is not due to an empty marker table but that none of the contigs in the markers table are present in other annotation tables (or vice versa).

There's a few scenarios that may cause this, feel free to add to the non-exhaustive list below:

1. The taxonomic assignment step resulted in misclassification of contigs
   • This could be due to database issues (no similar relatives, contamination, etc), miss-assembly of contigs, etc

To debug, you can use the following python code to check if there are any contigs in the marker table that are not in the other annotation tables:

import pandas as pd
from autometa.binning.utilities import read_annotations

kmers="~/kmers.tsv"
markers="~/bacteria.markers.tsv"
coverages="~/coverages.tsv"
gc_content="~/gc_content.tsv"
taxonomy="~/taxonomy.tsv"

main_df = read_annotations(
            [kmers, coverages, gc_content, taxonomy]
)
markers_df=pd.read_csv(markers,sep="\t",index_col=0)

main_df.loc[main_df.index.isin(markers_df.index)].empty

[chasemc]

Related to #203