"nan" values in Purity (bp) and Purity (seq)

[apcamargo]

Hi,

I'm using AMBER to evaluate a set of bins I obtained from a metagenome assembled from the CAMI Toy Mouse Gut Dataset reads. I've noticed that some bins have nan values in the Purity (bp) and Purity (seq) columns. What might be causing that?

To build the gold standard I aligned the reassembled contigs to the original genomes using BLAST, as described in Vamb's paper:

We removed any hits shorter than 500 bp or with lower nucleotide identity than 95%. If a query (reassembled) contig was aligned to multiple reference (original) contigs, we accepted the reference with the longest alignment, if the alignment was more than twice as long of the next longest. If that was not the case for any reference, we accepted the reference with highest nucleotide identity, if the reference was longer than 10 kbp, had an alignment length of at least 90% of the longest-aligning reference, and had at least 0.05% higher nucleotide identity than the second-highest identity reference. If no reference fit those criteria, they were ignored in the benchmarking.

Bin ID	Most abundant genome	Purity (bp)	Completeness (bp)	Bin size (bp)	True positives (bp)	True size of most abundant genome (bp)	Purity (seq)	Completeness (seq)	Bin size (seq)	True positives (seq)	True size of most abundant genome (seq)
mouse_gut_5.vamb.83	denovo8255.1	0.659	0.983	741631	488937	497562	0.655	0.980	226	148	151
mouse_gut_5.vamb.9	269125.1	0.998	0.977	1809513	1806528	1848507	0.994	0.977	172	171	175
mouse_gut_5.vamb.81	179513.0	1.000	0.963	807379	807379	838686	1.000	0.962	280	280	291
mouse_gut_5.vamb.126	228785.0	1.000	0.960	308660	308660	321520	1.000	0.954	103	103	108
mouse_gut_5.vamb.72	661259.1	1.000	0.940	241230	241230	256496	1.000	0.944	85	85	90
mouse_gut_5.vamb.182	259993.0	1.000	0.922	647949	647949	702765	1.000	0.917	211	211	230
mouse_gut_5.vamb.454	denovo11208.0	0.525	0.919	1065913	559173	608144	0.531	0.895	305	162	181
mouse_gut_5.vamb.11	133719.0	0.760	0.915	534069	405959	443893	0.748	0.888	127	95	107
mouse_gut_5.vamb.111	denovo11993.0	1.000	0.875	636280	636280	727174	1.000	0.882	194	194	220
mouse_gut_5.vamb.793	4471135.0	0.992	0.863	1913155	1898009	2200037	0.990	0.845	583	577	683
mouse_gut_5.vamb.333	denovo12532.0	nan	0.857	202574	199887	233197	nan	0.852	76	75	88
mouse_gut_5.vamb.51	denovo2465.0	nan	0.848	218613	218613	257907	nan	0.863	82	82	95
mouse_gut_5.vamb.115	denovo1032.0	nan	0.816	44891	24716	30297	nan	0.750	11	6	8
mouse_gut_5.vamb.71	denovo10679.0	0.333	0.787	654629	218217	277372	0.341	0.622	82	28	45
mouse_gut_5.vamb.451	denovo11206.0	0.893	0.761	1096925	979172	1287345	0.872	0.718	298	260	362
mouse_gut_5.vamb.428	denovo2609.0	0.998	0.733	1077877	1075782	1468385	0.997	0.717	325	324	452
mouse_gut_5.vamb.1036	263992.0	nan	0.231	66518	41954	181693	nan	0.242	23	15	62

[fernandomeyer]

Hello, the purity of the smallest bins is set to nan if you're using the --filter option, say, --filter 1. Then the smallest bins corresponding to 1% of the binned data by a binner is "removed", which in practice means that their purity is set to nan and they are not considered in the average purity.

[apcamargo]

Thanks!