Please follow the instructions provided in the README to install gbintk
.
Now we are all set for the demo.
Test data for the demo can be found in the tests/data/5G_metaSPAdes/
folder. Let's set the testing directory path to a variable as shown below so we don't have to type the path every time.
TESTDIR=tests/data/5G_metaSPAdes/
Run the following command to bin the test dataset using MetaCoAG.
gbintk metacoag --assembler spades --graph $TESTDIR/assembly_graph_with_scaffolds.gfa --contigs $TESTDIR/contigs.fasta --paths $TESTDIR/contigs.paths --abundance $TESTDIR/coverm_mean_coverage.tsv --output $TESTDIR
Run the following command to refine the binning results using GraphBin.
gbintk graphbin --assembler spades --graph $TESTDIR/assembly_graph_with_scaffolds.gfa --contigs $TESTDIR/contigs.fasta --paths $TESTDIR/contigs.paths --binned $TESTDIR/contig_to_bin.tsv --output $TESTDIR
Run the following command to visualise the original binning result and the refined binning results on the assembly graph.
gbintk visualise --assembler spades --initial $TESTDIR/contig_to_bin.tsv --final $TESTDIR/graphbin_output.csv --graph $TESTDIR/assembly_graph_with_scaffolds.gfa --paths $TESTDIR/contigs.paths --output $TESTDIR --width 2500 --height 2500
Below are some example visualisations generated.
Initial binning result
Refined binning result
Run the following command to evaluate the initial and final binning results.
gbintk evaluate --binned $TESTDIR/metacoag_res.csv --groundtruth $TESTDIR/ground_truth.csv --output $TESTDIR/
The output in evaluation_results.txt
will be as follows.
KxS Matrix:
-- -- - -- --
1 0 9 0 0
75 0 0 0 0
0 0 0 23 0
1 63 0 0 2
6 4 0 0 22
-- -- - -- --
Evaluation Results:
Precision = 93.20388349514563
Recall = 37.721021611001966
F1-score = 53.70629370629371
ARI = 84.17314460729499
gbintk evaluate --binned $TESTDIR/graphbin_res.csv --groundtruth $TESTDIR/ground_truth.csv --output $TESTDIR/
The output in evaluation_results.txt
will be as follows.
KxS Matrix:
-- --- --- -- ---
0 0 0 70 0
18 0 0 0 0
0 0 0 0 189
0 0 117 0 0
0 103 0 0 0
-- --- --- -- ---
Evaluation Results:
Precision = 100.0
Recall = 97.64243614931237
F1-score = 98.80715705765408
ARI = 100.0