Reads Binning Improves Alignment-Free Metagenome Comparison

POSMM: an efficient alignment-free metagenomic profiler that complements alignment-based profiling

We present here POSMM (pronounced 'Possum'), Python-Optimized Standard Markov Model classifier, which is a new incarnation of the Markov model approach to metagenomic sequence analysis. Built on the top of a rapid Markov model based classification algorithm SMM, POSMM reintroduces high sensitivity associated with alignment-free taxonomic classifiers to probe whole genome or metagenome datasets of increasingly prohibitive sizes. Logistic regression models generated and optimized using the Python sklearn library, transform Markov model probabilities to scores suitable for thresholding. Featuring a dynamic database-free approach, models are generated directly from genome fasta files per run, making POSMM a valuable accompaniment to many other programs. By combining POSMM with ultrafast classifiers such as Kraken2, their complementary strengths can be leveraged to produce higher overall accuracy in metagenomic sequence classification than by either as a standalone classifier. POSMM is a user-friendly and highly adaptable tool designed for broad use by the metagenome scientific community.

Phylogeny analysis of whole protein-coding genes in metagenomic data detected an environmental gradient for the microbiota

Environmental factors affect the growth of microorganisms and therefore alter the composition of microbiota. Correlative analysis of the relationship between metagenomic composition and the environmental gradient can help elucidate key environmental factors and establishment principles for microbial communities. However, a reasonable method to quantitatively compare whole metagenomic data and identify the primary environmental factors for the establishment of microbiota has not been reported so far. In this study, we developed a method to compare whole proteomes deduced from metagenomic shotgun sequencing data, and quantitatively display their phylogenetic relationships as metagenomic trees. We called this method Metagenomic Phylogeny by Average Sequence Similarity (MPASS). We also compared one of the metagenomic trees with dendrograms of environmental factors using a comparison tool for phylogenetic trees. The MPASS method correctly constructed metagenomic trees of simulated metagenomes and soil and water samples. The topology of the metagenomic tree of samples from the Kirishima hot springs area in Japan was highly similarity to that of the dendrograms based on previously reported environmental factors for this area. The topology of the metagenomic tree also reflected the dynamics of microbiota at the taxonomic and functional levels. Our results strongly suggest that MPASS can successfully classify metagenomic shotgun sequencing data based on the similarity of whole protein-coding sequences, and will be useful for the identification of principal environmental factors for the establishment of microbial communities. Custom Perl script for the MPASS pipeline is available at https://github.com/s0sat/MPASS.

Reads Binning Improves the Assembly of Viral Genome Sequences From Metagenomic Samples

Metagenomes can be considered as mixtures of viral, bacterial, and other eukaryotic DNA sequences. Mining viral sequences from metagenomes could shed insight into virus-host relationships and expand viral databases. Current alignment-based methods are unsuitable for identifying viral sequences from metagenome sequences because most assembled metagenomic contigs are short and possess few or no predicted genes, and most metagenomic viral genes are dissimilar to known viral genes. In this study, I developed a Markov model-based method, VirMC, to identify viral sequences from metagenomic data. VirMC uses Markov chains to model sequence signatures and construct a scoring model using a likelihood test to distinguish viral and bacterial sequences. Compared with the other two state-of-the-art viral sequence-prediction methods, VirFinder and PPR-Meta, my proposed method outperformed VirFinder and had similar performance with PPR-Meta for short contigs with length less than 400 bp. VirMC outperformed VirFinder and PPR-Meta for identifying viral sequences in contaminated metagenomic samples with eukaryotic sequences. VirMC showed better performance in assembling viral-genome sequences from metagenomic data (based on filtering potential bacterial reads). Applying VirMC to human gut metagenomes from healthy subjects and patients with type-2 diabetes (T2D) revealed that viral contigs could help classify healthy and diseased statuses. This alignment-free method complements gene-based alignment approaches and will significantly improve the precision of viral sequence identification.

geneRFinder: gene finding in distinct metagenomic data complexities

Background

Microbes perform a fundamental economic, social, and environmental role in our society. Metagenomics makes it possible to investigate microbes in their natural environments (the complex communities) and their interactions. The way they act is usually estimated by looking at the functions they play in those environments and their responsibility is measured by their genes. The advances of next-generation sequencing technology have facilitated metagenomics research however it also creates a heavy computational burden. Large and complex biological datasets are available as never before. There are many gene predictors available that can aid the gene annotation process though they lack handling appropriately metagenomic data complexities. There is no standard metagenomic benchmark data for gene prediction. Thus, gene predictors may inflate their results by obfuscating low false discovery rates.

Results

We introduce geneRFinder, an ML-based gene predictor able to outperform state-of-the-art gene prediction tools across this benchmark by using only one pre-trained Random Forest model. Average prediction rates of geneRFinder differed in percentage terms by 54% and 64%, respectively, against Prodigal and FragGeneScan while handling high complexity metagenomes. The specificity rate of geneRFinder had the largest distance against FragGeneScan, 79 percentage points, and 66 more than Prodigal. According to McNemar’s test, all percentual differences between predictors performances are statistically significant for all datasets with a 99% confidence interval.

Conclusions

We provide geneRFinder, an approach for gene prediction in distinct metagenomic complexities, available at gitlab.com/r.lorenna/generfinder and https://osf.io/w2yd6/, and also we provide a novel, comprehensive benchmark data for gene prediction—which is based on The Critical Assessment of Metagenome Interpretation (CAMI) challenge, and contains labeled data from gene regions—available at https://sourceforge.net/p/generfinder-benchmark.

Classifying the Lifestyle of Metagenomically-Derived Phages Sequences Using Alignment-Free Methods

Phages are viruses that infect bacteria. The phages can be classified into two different categories based on their lifestyles: temperate and lytic. Now, the metavirome can generate a large number of fragments from the viral genomic sequences of entire environmental community, which makes it impossible to determine their lifestyles through experiments. Thus, there is a need to development computational methods for annotating phage contigs and making prediction of their lifestyles. Alignment-based methods for classifying phage lifestyle are limited by incomplete assembled genomes and nucleotide databases. Alignment-free methods based on the frequencies of k-mers were widely used for genome and metagenome comparison which did not rely on the completeness of genome or nucleotide databases. To mimic fragmented metagenomic sequences, the temperate and lytic phages genomic sequences were split into non-overlapping fragments with different lengths, then, I comprehensively compared nine alignment-free dissimilarity measures with a wide range of choices of k-mer length and Markov orders for predicting the lifestyles of these phage contigs. The dissimilarity measure, d2S, performed better than other dissimilarity measures for classifying the lifestyles of phages. Thus, I propose that the alignment-free method, d2S, can be used for predicting the lifestyles of phages which derived from the metagenomic data.

KmerGO: A Tool to Identify Group-Specific Sequences With k-mers

Capturing group-specific sequences between two groups of genomic/metagenomic sequences is critical for the follow-up identifications of singular nucleotide variants (SNVs), gene families, microbial species or other elements associated with each group. A sequence that is present, or rich, in one group, but absent, or scarce, in another group is considered a “group-specific” sequence in our study. We developed a user-friendly tool, KmerGO, to identify group-specific sequences between two groups of genomic/metagenomic long sequences or high-throughput sequencing datasets. Compared with other tools, KmerGO captures group-specific k-mers (k up to 40 bps) with much lower requirements for computing resources in much shorter running time. For a 1.05 TB dataset (.fasta), it takes KmerGO about 21.5 h (including k-mer counting) to return assembled group-specific sequences on a regular stand-alone workstation with no more than 1 GB memory. Furthermore, KmerGO can also be applied to capture trait-associated sequences for continuous-trait. Through multi-process parallel computing, KmerGO is implemented with both graphic user interface and command line on Linux and Windows free from any pre-installed supporting environments, packages, and complex configurations. The output group-specific k-mers or sequences from KmerGO could be the inputs of other tools for the downstream discovery of biomarkers, such as genetic variants, species, or genes. KmerGO is available at https://github.com/ChnMasterOG/KmerGO.

Binning unassembled short reads based on k-mer abundance covariance using sparse coding

BACKGROUND

Sequence-binning techniques enable the recovery of an increasing number of genomes from complex microbial metagenomes and typically require prior metagenome assembly, incurring the computational cost and drawbacks of the latter, e.g., biases against low-abundance genomes and inability to conveniently assemble multi-terabyte datasets.

RESULTS

We present here a scalable pre-assembly binning scheme (i.e., operating on unassembled short reads) enabling latent genome recovery by leveraging sparse dictionary learning and elastic-net regularization, and its use to recover hundreds of metagenome-assembled genomes, including very low-abundance genomes, from a joint analysis of microbiomes from the LifeLines DEEP population cohort (n = 1,135, >1010 reads).

CONCLUSION

We showed that sparse coding techniques can be leveraged to carry out read-level binning at large scale and that, despite lower genome reconstruction yields compared to assembly-based approaches, bin-first strategies can complement the more widely used assembly-first protocols by targeting distinct genome segregation profiles. Read enrichment levels across 6 orders of magnitude in relative abundance were observed, indicating that the method has the power to recover genomes consistently segregating at low levels.