Metagenomic binning with assembly graph embeddings

Abstract

Motivation Despite recent advancements in sequencing technologies and assembly methods, obtaining high-quality microbial genomes from metagenomic samples is still not a trivial task. Current metagenomic binners do not take full advantage of assembly graphs and are not optimized for long-read assemblies. Deep graph learning algorithms have been proposed in other fields to deal with complex graph data structures. The graph structure generated during the assembly process could be integrated with contig features to obtain better bins with deep learning. Results We propose GraphMB, which uses graph neural networks to incorporate the assembly graph into the binning process. We test GraphMB on long-read datasets of different complexities, and compare the performance with other binners in terms of the number of High Quality (HQ) genome bins obtained. With our approach, we were able to obtain unique bins on all real datasets, and obtain more bins on most datasets. In particular, we obtained on average 17.5% more HQ bins when compared to state-of-the-art binners and 13.7% when aggregating the results of our binner with the others. These results indicate that a deep learning model can integrate contig-specific and graph-structure information to improve metagenomic binning. Availability GraphMB is available from https://github.com/MicrobialDarkMatter/GraphMB

SemiBin2: self-supervised contrastive learning leads to better MAGs for short- and long-read sequencing

“…For the long-read datasets, we compared to MetaBAT2 16 , MetaDecoder 18 , VAMB 19 , and SemiBin1 20 . We did not include GraphMB 25 and LRBinner 24 in this comparison because we used gold standard assemblies for binning in simulated datasets, and we could not obtain the assembly graph GraphMB requires as input and LRBinner cannot be run with co-assembly binning.…”

“…SemiBin2 uses the implementation of DBSCAN in scikit-learn 34 and runs DBSCAN with ε value equals to 0.01, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, 0.5, and 0.55. Then SemiBin2 integrates the results of these runs based on the single-copy genes that have been used in other tools 18,25 . In particular, SemiBin2 uses 107 single-copy genes 32 to estimate the completeness, contamination, and F1-score of every bin.…”

“…For short-read datasets with multi-sample binning, we compared to VAMB (version 3.0.7, -m 2000 ) 19 and SemiBin1 (version 1.0.0) 20 , existing binners supporting for multi-sample binning. For long-read datasets, we compared to MetaBAT2 (version 2.15, –percentIdentity 50 ) 16 , VAMB (version 3.0.7, -m 2000 ) 19 , SemiBin1 (version 1.0.0, –environment human_gut for PR-JCA007414) 20 , GraphMB (version 0.1.4) 25 , MetaDecoder (version 1.0.13) 18 and LRBinner (version 2.1) 24 .…”

“…Thus, SemiBin2 proposes an ensemble-based DBSCAN clustering algorithm to extend to long-read data. We compared it to the other binners proposed for long-read sequencing data, such as LRBinner 23,24 and the recently-proposed GraphMB 25 , and show that it outperforms them.…”

Comparison between ribosomal assembly and machine learning tools for microbial identification of organisms with different characteristics

“…Woloszynek et al explore using word and sentence embedding approaches for nucleotide sequences, which may be a suitable numerical representation for downstream machine learning applications (especially deep learning). The results show that embedding sequences results in meaningful representations that can be used for exploratory analyses or for downstream applications, such as binning [31][32][33]. Choi et al present all-to-all comparison of metagenomes using k-mer content and Hadoop for precise clustering [34].…”