K-mer Size Research Articles

Strain-resolved de-novo metagenomic assembly of viral genomes and microbial 16S rRNAs

BackgroundMetagenomics is a powerful approach to study environmental and human-associated microbial communities and, in particular, the role of viruses in shaping them. Viral genomes are challenging to assemble from metagenomic samples due to their genomic diversity caused by high mutation rates. In the standard de Bruijn graph assemblers, this genomic diversity leads to complex k-mer assembly graphs with a plethora of loops and bulges that are challenging to resolve into strains or haplotypes because variants more than the k-mer size apart cannot be phased. In contrast, overlap assemblers can phase variants as long as they are covered by a single read.ResultsHere, we present PenguiN, a software for strain resolved assembly of viral DNA and RNA genomes and bacterial 16S rRNA from shotgun metagenomics. Its exhaustive detection of all read overlaps in linear time combined with a Bayesian model to select strain-resolved extensions allow it to assemble longer viral contigs, less fragmented genomes, and more strains than existing assembly tools, on both real and simulated datasets. We show a 3–40-fold increase in complete viral genomes and a 6-fold increase in bacterial 16S rRNA genes.ConclusionPenguiN is the first overlap-based assembler for viral genome and 16S rRNA assembly from large and complex metagenomic datasets, which we hope will facilitate studying the key roles of viruses in microbial communities.C22dm7AjjRZTtiuc33W1snVideo

Prokrustean Graph: A substring index for rapid k-mer size analysis.

Despite the widespread adoption of -mer-based methods in bioinformatics, understanding the influence of -mer sizes remains a persistent challenge. Selecting an optimal -mer size or employing multiple -mer sizes is often arbitrary, application-specific, and fraught with computational complexities. Typically, the influence of -mer size is obscured by the outputs of complex bioinformatics tasks, such as genome analysis, comparison, assembly, alignment, and error correction. However, it is frequently overlooked that every method is built above a well-defined -mer-based object like Jaccard Similarity, de Bruijn graphs, -mer spectra, and Bray-Curtis Dissimilarity. Despite these objects offering a clearer perspective on the role of -mer sizes, the dynamics of -mer-based objects with respect to -mer sizes remain surprisingly elusive. This paper introduces a computational framework that generalizes the transition of -mer-based objects across -mer sizes, utilizing a novel substring index, the Prokrustean graph. The primary contribution of this framework is to compute quantities associated with -mer-based objects for all -mer sizes, where the computational complexity depends solely on the number of maximal repeats and is independent of the range of -mer sizes. For example, counting vertices of compacted de Bruijn graphs for can be accomplished in mere seconds with our substring index constructed on a gigabase-sized read set. Additionally, we derive a space-efficient algorithm to extract the Prokrustean graph from the Burrows-Wheeler Transform. It becomes evident that modern substring indices, mostly based on longest common prefixes of suffix arrays, inherently face difficulties at exploring varying -mer sizes due to their limitations at grouping co-occurring substrings. We have implemented four applications that utilize quantities critical in modern pangenomics and metagenomics. The code for these applications and the construction algorithm is available at https://github.com/KoslickiLab/prokrustean.

Enhanced Compression of k-Mer Sets with Counters via de Bruijn Graphs.

An essential task in computational genomics involves transforming input sequences into their constituent k-mers. The quest for an efficient representation of k-mer sets is crucial for enhancing the scalability of bioinformatic analyses. One widely used method involves converting the k-mer set into a de Bruijn graph (dBG), followed by seeking a compact graph representation via the smallest path cover. This study introduces USTAR* (Unitig STitch Advanced constRuction), a tool designed to compress both a set of k-mers and their associated counts. USTAR leverages the connectivity and density of dBGs, enabling a more efficient path selection for constructing the path cover. The efficacy of USTAR is demonstrated through its application in compressing real read data sets. USTAR improves the compression achieved by UST (Unitig STitch), the best algorithm, by percentages ranging from 2.3% to 26.4%, depending on the k-mer size, and it is up to times faster.

Experimental and analytical pipeline for sub-genomic RNA landscape of coronavirus by Nanopore sequencer.

Coronaviruses (CoVs), including severe acute respiratory syndrome coronavirus 2, can infect a variety of mammalian and avian hosts with significant medical and economic consequences. During the life cycle of CoV, a coordinated series of subgenomic RNAs, including canonical subgenomic messenger RNA and non-canonical defective viral genomes (DVGs), are generated with different biological implications. Studies that adopted the Nanopore sequencer (ONT) to investigate the landscape and dynamics of viral RNA subgenomic transcriptomes applied arbitrary bioinformatics parameters without justification or experimental validation. The current study used bovine coronavirus (BCoV), which can be performed under biosafety level 2 for library construction and experimental validation using traditional colony polymerase chain reaction and Sanger sequencing. Four different ONT protocols, including RNA direct and cDNA direct sequencing with or without exonuclease treatment, were used to generate RNA transcriptomic libraries from BCoV-infected cell lysates. Through rigorously examining the k-mer, gap size, segment size, and bin size, the optimal cutoffs for the bioinformatic pipeline were determined to remove the sequence noise while keeping the informative DVG reads. The sensitivity and specificity of identifying DVG reads using the proposed pipeline can reach 82.6% and 99.6% under the k-mer size cutoff of 15. Exonuclease treatment reduced the abundance of RNA transcripts; however, it was not necessary for future library preparation. Additional recovery of clipped BCoV nucleotide sequences with experimental validation expands the landscape of the CoV discontinuous RNA transcriptome, whose biological function requires future investigation. The results of this study provide the benchmarks for library construction and bioinformatic parameters for studying the discontinuous CoV RNA transcriptome.IMPORTANCEFunctional defective viral genomic RNA, containing all the cis-acting elements required for translation or replication, may play different roles in triggering cell innate immune signaling, interfering with the canonical subgenomic messenger RNA transcription/translation or assisting in establishing persistence infection. This study does not only provide benchmarks for library construction and bioinformatic parameters for studying the discontinuous coronavirus RNA transcriptome but also reveals the complexity of the bovine coronavirus transcriptome, whose functional assays will be critical in future studies.

Protecting Genetic Genealogical Databases from Identical-by-State Probing Attacks: A Machine Learning-Based Approach

Identical-by-state (IBS) probing is a way of attacking a public genealogy database to discover the identities of people who share specific qualities. The attacker creates an IBS-inert DNA sequence (IBS-inert-DNA-sequence) and combines it with a sequence containing the trait of interest. The final sequence matches people with genomic areas similar to the trait. To prevent attacks, it was hypothesized that the design of IBS-inert-DNA-sequences is based on the principle that they are susceptible to detection by skilled machine learning systems, because the attacker purposely creates an IBS-inert-DNA-sequence which is structurally dissimilar to real DNA sequences. The dataset consisted of real DNA (from the UCI Machine Learning Repository's splice junction gene sequences dataset) and computer-generated sequences. Eighteen non-identical Random Forest classifier models (RF-models) were created to determine the best configurations for discriminating between real and computer-generated sequences. The findings revealed that an optimized RF-model combined with k-mer and n-gram values of 2 each, resulted in the most performant model, with accuracy, sensitivity, specificity, false positive rate, Matthew’s correlation coefficient, and area under the receiver operating characteristic curve values of 88.3%, 84.8%, 91.8%, 8.2%, 0.768 and 0.958, respectively. A decline in performance was linked to an increase in k-mer size.

Minimal encodings of canonical k-mers for general alphabets and even k-mer sizes

Minimal encodings of canonical k-mers for general alphabets and even k-mer sizes

Proving sequence aligners can guarantee accuracy in almost O(m log n) time through an average-case analysis of the seed-chain-extend heuristic.

Seed-chain-extend with k-mer seeds is a powerful heuristic technique for sequence alignment used by modern sequence aligners. Although effective in practice for both runtime and accuracy, theoretical guarantees on the resulting alignment do not exist for seed-chain-extend. In this work, we give the first rigorous bounds for the efficacy of seed-chain-extend with k-mers in expectation Assume we are given a random nucleotide sequence of length ∼n that is indexed (or seeded) and a mutated substring of length ∼m ≤ n with mutation rate θ < 0.206. We prove that we can find a k = Θ(log n) for the k-mer size such that the expected runtime of seed-chain-extend under optimal linear-gap cost chaining and quadratic time gap extension is O(mn f (θ) log n), where f(θ) < 2.43 · θ holds as a loose bound. The alignment also turns out to be good; we prove that more than [Formula: see text] fraction of the homologous bases is recoverable under an optimal chain. We also show that our bounds work when k-mers are sketched, that is, only a subset of all k-mers is selected, and that sketching reduces chaining time without increasing alignment time or decreasing accuracy too much, justifying the effectiveness of sketching as a practical speedup in sequence alignment. We verify our results in simulation and on real noisy long-read data and show that our theoretical runtimes can predict real runtimes accurately. We conjecture that our bounds can be improved further, and in particular, f(θ) can be further reduced.

ASmiR: a machine learning framework for prediction of abiotic stress-specific miRNAs in plants.

Abiotic stresses have become a major challenge in recent years due to their pervasive nature and shocking impacts on plant growth, development, and quality. MicroRNAs (miRNAs) play a significant role in plant response to different abiotic stresses. Thus, identification of specific abiotic stress-responsive miRNAs holds immense importance in crop breeding programmes to develop cultivars resistant to abiotic stresses. In this study, we developed a machine learning-based computational model for prediction of miRNAs associated with four specific abiotic stresses such as cold, drought, heat and salt. The pseudo K-tuple nucleotide compositional features of Kmer size 1 to 5 were used to represent miRNAs in numeric form. Feature selection strategy was employed to select important features. With the selected feature sets, support vector machine (SVM) achieved the highest cross-validation accuracy in all four abiotic stress conditions. The highest cross-validated prediction accuracies in terms of area under precision-recall curve were found to be 90.15, 90.09, 87.71, and 89.25% for cold, drought, heat and salt respectively. Overall prediction accuracies for the independent dataset were respectively observed 84.57, 80.62, 80.38 and 82.78%, for the abiotic stresses. The SVM was also seen to outperform different deep learning models for prediction of abiotic stress-responsive miRNAs. To implement our method with ease, an online prediction server "ASmiR" has been established at https://iasri-sg.icar.gov.in/asmir/ . The proposed computational model and the developed prediction tool are believed to supplement the existing effort for identification of specific abiotic stress-responsive miRNAs in plants.

CNN_FunBar: Advanced Learning Technique for Fungi ITS Region Classification.

Fungal species identification from metagenomic data is a highly challenging task. Internal Transcribed Spacer (ITS) region is a potential DNA marker for fungi taxonomy prediction. Computational approaches, especially deep learning algorithms, are highly efficient for better pattern recognition and classification of large datasets compared to in silico techniques such as BLAST and machine learning methods. Here in this study, we present CNN_FunBar, a convolutional neural network-based approach for the classification of fungi ITS sequences from UNITE+INSDC reference datasets. Effects of convolution kernel size, filter numbers, k-mer size, degree of diversity and category-wise frequency of ITS sequences on classification performances of CNN models have been assessed at all taxonomic levels (species, genus, family, order, class and phylum). It is observed that CNN models can produce >93% average accuracy for classifying ITS sequences from balanced datasets with 500 sequences per category and 6-mer frequency features at all levels. The comparative study has revealed that CNN_FunBar can outperform machine learning-based algorithms (SVM, KNN, Naïve-Bayes and Random Forest) as well as existing fungal taxonomy prediction software (funbarRF, Mothur, RDP Classifier and SINTAX). The present study will be helpful for fungal taxonomy classification using large metagenomic datasets.

CMash: fast, multi-resolution estimation of k-mer-based Jaccard and containment indices.

Motivation K-mer-based methods are used ubiquitously in the field of computational biology. However, determining the optimal value of k for a specific application often remains heuristic. Simply reconstructing a new k-mer set with another k-mer size is computationally expensive, especially in metagenomic analysis where datasets are large. Here, we introduce a hashing-based technique that leverages a kind of bottom-m sketch as well as a k-mer ternary search tree (KTST) to obtain k-mer-based similarity estimates for a range of k values. By truncating k-mers stored in a pre-built KTST with a large value, we can simultaneously obtain k-mer-based estimates for all k values up to kmax. This truncation approach circumvents the reconstruction of new k-mer sets when changing k values, making analysis more time and space-efficient.ResultsWe derived the theoretical expression of the bias factor due to truncation. And we showed that the biases are negligible in practice: when using a KTST to estimate the containment index between a RefSeq-based microbial reference database and simulated metagenome data for 10 values of k, the running time was close to 10× faster compared to a classic MinHash approach while using less than one-fifth the space to store the data structure.Availability and implementationA python implementation of this method, CMash, is available at https://github.com/dkoslicki/CMash. The reproduction of all experiments presented herein can be accessed via https://github.com/KoslickiLab/CMASH-reproducibles.Supplementary information Supplementary data are available at Bioinformatics online.

Random adsorption process of linear k-mers on square lattices under the Achlioptas process.

We study the explosive percolation with k-mer random sequential adsorption (RSA) process. We consider both the Achlioptas process (AP) and the inverse Achlioptas process (IAP), in which giant cluster formation is prohibited and accelerated, respectively. By employing finite-size scaling analysis, we confirm that the percolation transitions are continuous, and thus we calculate the percolation threshold and critical exponents. This allows us to determine the universality class of the k-mer explosive percolation transition. Interestingly, the numerical simulation suggests that the universality class of the explosive percolation transition with the AP alters when the k-mer size changes. In contrast, the universality class of the transition with the IAP is independent of k, but it differs from that of the RSA without the IAP.

Multiplex de Bruijn graphs enable genome assembly from long, high-fidelity reads.

Although most existing genome assemblers are based on de Bruijn graphs, the construction of these graphs for large genomes and large k-mer sizes has remained elusive. This algorithmic challenge has become particularly pressing with the emergence of long, high-fidelity (HiFi) reads that have been recently used to generate a semi-manual telomere-to-telomere assembly of the human genome. To enable automated assemblies of long, HiFi reads, we present the La Jolla Assembler (LJA), a fast algorithm using the Bloom filter, sparse de Bruijn graphs and disjointig generation. LJA reduces the error rate in HiFi reads by three orders of magnitude, constructs the de Bruijn graph for large genomes and large k-mer sizes and transforms it into a multiplex de Bruijn graph with varying k-mer sizes. Compared to state-of-the-art assemblers, our algorithm not only achieves five-fold fewer misassemblies but also generates more contiguous assemblies. We demonstrate the utility of LJA via the automated assembly of a human genome that completely assembled six chromosomes.

Lerna: transformer architectures for configuring error correction tools for short- and long-read genome sequencing

BackgroundSequencing technologies are prone to errors, making error correction (EC) necessary for downstream applications. EC tools need to be manually configured for optimal performance. We find that the optimal parameters (e.g., k-mer size) are both tool- and dataset-dependent. Moreover, evaluating the performance (i.e., Alignment-rate or Gain) of a given tool usually relies on a reference genome, but quality reference genomes are not always available. We introduce Lerna for the automated configuration of k-mer-based EC tools. Lerna first creates a language model (LM) of the uncorrected genomic reads, and then, based on this LM, calculates a metric called the perplexity metric to evaluate the corrected reads for different parameter choices. Next, it finds the one that produces the highest alignment rate without using a reference genome. The fundamental intuition of our approach is that the perplexity metric is inversely correlated with the quality of the assembly after error correction. Therefore, Lerna leverages the perplexity metric for automated tuning of k-mer sizes without needing a reference genome.ResultsFirst, we show that the best k-mer value can vary for different datasets, even for the same EC tool. This motivates our design that automates k-mer size selection without using a reference genome. Second, we show the gains of our LM using its component attention-based transformers. We show the model’s estimation of the perplexity metric before and after error correction. The lower the perplexity after correction, the better the k-mer size. We also show that the alignment rate and assembly quality computed for the corrected reads are strongly negatively correlated with the perplexity, enabling the automated selection of k-mer values for better error correction, and hence, improved assembly quality. We validate our approach on both short and long reads. Additionally, we show that our attention-based models have significant runtime improvement for the entire pipeline—18times faster than previous works, due to parallelizing the attention mechanism and the use of JIT compilation for GPU inferencing.ConclusionLerna improves de novo genome assembly by optimizing EC tools. Our code is made available in a public repository at: https://github.com/icanforce/lerna-genomics.

Controlling local relaxation in small clusters of magnetic nanoparticles

We investigate the local (particle level) and averaged magnetic relaxation characteristics in linear chain-like agglomerates of k magnetic nanoparticles (MNPs) or k-mers using computer simulations. The local relaxation behaviour is dictated by the corresponding dipolar field acting on the nanoparticle, irrespective of k-mer size. There is a wide variation in local relaxation characteristics as a function of nanoparticle position in a small nanocluster. On the other hand, there is more uniformity in the local relaxation response with a larger k-mer, except for MNPs at the boundary. Interestingly, there is a smooth decay of magnetization with small hd, independent of cluster size. In contrast, the magnetization ceases to relax in the presence of substantial dipolar interaction strength. Remarkably, the local Néel relaxation time τNi is directly proportional to the corresponding dipolar field. Likewise, the averaged Néel relaxation time τN also depends strongly on the k-mer size and hd.

1248. A Machine-Learning Approach to Predict the Cefazolin Inoculum Effect in Methicillin-Susceptible Staphylococcus aureus

Abstract Background The cefazolin (Cz) inoculum effect (CzIE), defined as an increase in the Cz MIC to ≥16 µg/mL at high inoculum (107 CFU/mL), has been associated with poor outcomes in MSSA bacteremia and osteomyelitis. The CzIE is associated with the BlaZ β-lactamase, encoded by blaZ and regulated by BlaR (antibiotic sensor) and BlaI (transcriptional repressor). Here, we aimed to obtain a machine-learning (ML) model to predict the presence of the CzIE based on the nucleotide sequence of the entire bla operon and its regulatory components. Methods Using whole genome sequencing, we analyzed the nucleotide sequences of the entire bla operon in 436 MSSA isolates recovered from blood, soft-tissue infections or pneumonia in adults (training-testing cohort, prevalence of the CzIE: 46%). Also, 32 MSSA recovered from pediatric patients with osteomyelitis with the CzIE were included as validation cohort. The CzIE was determined by broth microdilution at high inoculum. K-mer counts were obtained from the bla operon sequences of the isolates from the testing-training cohort, and then used in a ML pipeline which i) discards uninformative K-mers, ii) identifies optimal hyper-parameters and, iii) performs training of the model using 70% of the sequences as training set and 30% as testing set. The pipeline tested 11 different K-mer sizes and 2 models: Logistic Regression (LR) and Support Vector Machine (SVM). Finally, the model with best predictive ability was applied to the sequences of the MSSA osteomyelitis isolates (validation cohort). Results The ML approach had high specificity ( &amp;gt;90%), accuracy ( &amp;gt;80%) and ROC-AUC values ( &amp;gt;0.7) for detecting the CzIE in the testing set of isolates (Figure 1), independently of the type of model or the K-mer size used. The best predictive ability was with LR using K-mers of 17 nucleotides, with an accuracy of 84%, specificity of 96%, and sensitivity of 70% in the testing set (Figure 2). In the validation cohort, the model was capable to correctly identify all the strains exhibiting the CzIE (100% sensitivity). Figure 1. Prediction metrics of the ML pipeline for the detection of the CzIE in MSSA isolates from the training-test cohort. Predictions are shown accordingly to the model and K-mer sizes tested. Figure 2. ROC of best predictive model (Logistic Regression, K-mer size 17) for the detection of the CzIE in MSSA isolates. Conclusion The ML approach is a promising genomic application to detect the CzIE in MSSA isolates of a variety of sources, bypassing phenotypic testing. Further validation is needed to evaluate its possible utility in clinical settings. Disclosures Jonathon C. McNeil, MD, Agency for Healthcare Research and Quality (Research Grant or Support)Allergan (Grant/Research Support)Nabriva (Grant/Research Support, Other Financial or Material Support, Site PI for a multicenter trial) Anthony R. Flores, MD, MPH, PhD, Nothing to disclose Sheldon L. Kaplan, MD, Pfizer (Research Grant or Support) Cesar A. Arias, M.D., MSc, Ph.D., FIDSA, Entasis Therapeutics (Grant/Research Support)MeMed Diagnostics (Grant/Research Support)Merk (Grant/Research Support) Lorena Diaz, PhD , Nothing to disclose

A novel feature extraction method based on highly expressed SNPs for tissue-specific gene prediction

BackgroundGene expression provides a means for an organism to produce gene products necessary for the organism to live. Variation in the significant gene expression levels can distinguish the gene and the tissue in which the gene is expressed. Tissue-specific gene expression, often determined by single nucleotide polymorphisms (SNPs), provides potential molecular markers or therapeutic targets for disease progression. Therefore, SNPs are good candidates for identifying disease progression. The current bioinformatics literature uses gene network modeling to summarize complex interactions between transcription factors, genes, and gene products. Here, our focus is on the SNPs’ impact on tissue-specific gene expression levels. To the best of our knowledge, we are not aware of any studies that distinguish tissue-specific genes using SNP expression levels.MethodWe propose a novel feature extraction method based on highly expressed SNPs using k-mers as features. We also propose optimal k-mer and feature sizes used in our approach. Determining the optimal sizes is still an open research question as it depends on the dataset and purpose of the analysis. Therefore, we evaluate our algorithm’s performance on a range of k-mer and feature sizes using a multinomial naive Bayes (MNB) classifier on genes in the 49 human tissues from the Genotype-Tissue Expression (GTEx) portal.ConclusionsOur approach achieves practical performance results with k-mers of size 3. Based on the purpose of the analysis and the number of tissue-specific genes under study, feature sizes [7, 8, 9] and [8, 9, 10] are typically optimal for the machine learning model.

Pincho: A Modular Approach to High Quality De Novo Transcriptomics.

Transcriptomic reconstructions without reference (i.e., de novo) are common for data samples derived from non-model biological systems. These assemblies involve massive parallel short read sequence reconstructions from experiments, but they usually employ ad-hoc bioinformatic workflows that exhibit limited standardization and customization. The increasing number of transcriptome assembly software continues to provide little room for standardization which is exacerbated by the lack of studies on modularity that compare the effects of assembler synergy. We developed a customizable management workflow for de novo transcriptomics that includes modular units for short read cleaning, assembly, validation, annotation, and expression analysis by connecting twenty-five individual bioinformatic tools. With our software tool, we were able to compare the assessment scores based on 129 distinct single-, bi- and tri-assembler combinations with diverse k-mer size selections. Our results demonstrate a drastic increase in the quality of transcriptome assemblies with bi- and tri- assembler combinations. We aim for our software to improve de novo transcriptome reconstructions for the ever-growing landscape of RNA-seq data derived from non-model systems. We offer guidance to ensure the most complete transcriptomic reconstructions via the inclusion of modular multi-assembly software controlled from a single master console.

The impact of codon choice on translation process in Saccharomyces cerevisiae: folding class, protein function and secondary structure

The genetic code consists in a set of rules used by living organisms to translate genomic information, contained in genes, into proteins; every amino acid is coded by a set of nucleotide triplets or codons. We refer to codon choice as the choice of a given codon, among the synonymous available ones, to code a given amino acid occurrence.The aim of this work is to shed light on the pivotal role that codon choice plays in regulating the timing of translation process, through patterns of low and high translation efficiency codons. A translation efficiency value, namely codon score, was associated to each codon through a formula based on the number of tRNAs gene copies able to translate the given codon. By using codon scores, those k-mers of the proteome of Saccharomyces cerevisiae, showing low and high average scores associated to the correspondent codons, were computed.The analysis of distribution of both low and high average score k-mers clearly showed that, in particular for higher k-mer size, they occur much more than expected, strongly suggesting a functional role. Moreover performed analysis highlighted that significant k-mers preferentially occur in some protein folding classes, such as those containing alpha helices, and in some functional classes mainly involved in transcription process while codon choice seems to have a very low impact in proteins associated to energy production and metabolism. The relationship between secondary structures and significant k-mers was investigated, revealing that low score k-mers tend to preferentially occur in coil or close to coil regions and almost never in beta sheets, while high score k-mers preferentially occur in alpha helices, avoiding beta sheets, and close to coil regions for high k-mer sizes.Finally the analysis of distribution of significant codon patterns along the proteins highlighted a relevant enrichment of low average score k-mers at the 5’ end of protein-coding sequences in the region from 5th to 25th amino acid.

Taxonomic classification of metagenomic sequences from Relative Abundance Index profiles using deep learning

We propose a Convolutional Neural Network approach based on k-mer representation for metagenomic fragment classification problem. The proposed model consists of two steps; the first step is representation of DNA based on k-mer frequency with Relative Abundance Index (RAI) and the second step is classification metagenomic fragments with CNN. RAI scores, as DNA fragment representations are fed to CNN classifiers (CNN-RAI). RAI consist of the over- and under abundance statistics gathered from the taxon for each k-mer. In order to compare the performances of CNN-RAI and RAIphy, which classifies metagenomic fragments using the same input attributes with an expectation-maximization based approach, databases of different metagenomic scenarios were tested. Metagenomics data that were generated (or simulated) by different Next-Generation Sequencing platforms, respectively Illumina technology and Oxford Nanopore MinION were compiled into shotgun metagenomics or 16S rRNA datasets. RAI based method and CNN models were trained on represented data with read lengths ranging between 200 and 10,000 bp, also with distinct k-mer size (3≤k≤7) at genus level. RAI score was used for the first time in the deep learning algorithm as a spectral representation with improved performance thanks to the ability of deep learning on each dataset for a range of parameters. The proposed representation was compared to the current spectral methods and shown to be competitive for all datasets used in this study.

Comparison of machine learning and deep learning techniques in promoter prediction across diverse species.

Gene promoters are the key DNA regulatory elements positioned around the transcription start sites and are responsible for regulating gene transcription process. Various alignment-based, signal-based and content-based approaches are reported for the prediction of promoters. However, since all promoter sequences do not show explicit features, the prediction performance of these techniques is poor. Therefore, many machine learning and deep learning models have been proposed for promoter prediction. In this work, we studied methods for vector encoding and promoter classification using genome sequences of three distinct higher eukaryotes viz. yeast (Saccharomyces cerevisiae), A. thaliana (plant) and human (Homo sapiens). We compared one-hot vector encoding method with frequency-based tokenization (FBT) for data pre-processing on 1-D Convolutional Neural Network (CNN) model. We found that FBT gives a shorter input dimension reducing the training time without affecting the sensitivity and specificity of classification. We employed the deep learning techniques, mainly CNN and recurrent neural network with Long Short Term Memory (LSTM) and random forest (RF) classifier for promoter classification at k-mer sizes of 2, 4 and 8. We found CNN to be superior in classification of promoters from non-promoter sequences (binary classification) as well as species-specific classification of promoter sequences (multiclass classification). In summary, the contribution of this work lies in the use of synthetic shuffled negative dataset and frequency-based tokenization for pre-processing. This study provides a comprehensive and generic framework for classification tasks in genomic applications and can be extended to various classification problems.