Prediction of biological components in samples based on machine learning

Abstract

Methods and systems are provided that include training a machine learning model for detecting a biological component in a sample. A computer-implemented method and a system for performing the method may include collecting metagenomic data including biological components obtained from a sample; generating a first molecular data set of covariables; generating a second molecular data set of covariables; generating a training set comprising the first covariable and the second covariable and a combined covariable; and training the machine learning model using the polymeric molecular covariants to predict biological components from metagenomic data.

Description

[0001] Incorporation by reference into any priority application

[0002] This application claims priority to U.S. Provisional Application No. 63 / 502,393, filed on May 15, 2023, and U.S. Provisional Application No. 63 / 505,366, filed on May 31, 2023, the contents of each of which are incorporated by reference in their entirety. Background Art

[0003] Metagenomics, i.e., genomic analysis of microbial populations, makes it possible to profile microbial communities in the environment and the human body with unprecedented depth and breadth. Its rapidly expanding applications are revolutionizing our understanding of microbial diversity in natural and artificial environments, and linking microbial community profiles to health and disease. To date, most studies have relied on PCR amplification of microbial marker genes (e.g., bacterial 16S rRNA), for which large, selected databases have been established. Recently, higher throughput and lower-cost sequencing technologies have made it possible to turn to methods (both targeted and target-independent) for direct metagenomic analysis from specimens. These methods can reduce bias because they do not involve PCR primer binding, improve the resolution of genetically related taxa, and enable the discovery of new pathogens.

[0004] While conventional pathogen-specific nucleic acid amplification tests are highly sensitive and specific, they often require prior knowledge of likely pathogens. The result is an increasingly large but essentially limited diagnostic gene panel that enables diagnosis of the most common pathogens. In contrast, target-independent high-throughput sequencing allows for unbiased, hypothesis-free detection and molecular typing of a theoretically unlimited number of common and uncommon pathogens. The widespread availability of next-generation sequencing instruments, lower reagent costs, and simplified sample preparation protocols have enabled an increasing number of researchers to perform high-throughput DNA and RNA-seq for metagenomic studies. However, analysis of sequencing data remains difficult and time-consuming, requiring bioinformatics skills, computational resources, and microbiology expertise that many laboratories (particularly diagnostic laboratories) do not possess. Summary of the invention

[0005] The methods and systems disclosed herein each have several aspects, no one of which is solely responsible for its desired properties. Without limiting the scope of the claims, some salient features will now be briefly discussed. Many other embodiments are also contemplated, including embodiments with fewer, additional and / or different components, steps, features, purposes, benefits and advantages. The various components, aspects and steps may also be arranged and ordered in different ways. After considering this discussion, and particularly after reading the section entitled "Specific Embodiments," it will be understood how the features of the apparatus and methods disclosed herein provide advantages over other known apparatus and methods.

[0006] A computer-implemented method for training a machine learning model for detecting biological components in a sample, the method comprising collecting metagenomic data from biological components in the sample; generating a first molecular data set; generating a second molecular data set; creating a training set comprising an aggregated set of the first molecular data set and the second molecular data set; and training the machine learning model using the training set. In some embodiments, the machine learning model comprises a random forest model. In some embodiments, the machine learning model comprises a deep neural network (DNN). In some embodiments, the machine learning model comprises a convolutional neural network (CNN). In some embodiments, the machine learning model comprises a support vector machine (SVM). In some embodiments, the machine learning model comprises a category classifier.

[0007] In some embodiments, the method further comprises selecting a first machine learning model based on one or more metrics of the first molecular data set. In some embodiments, the method further comprises selecting a second machine learning model based on one or more metrics of the second molecular data set. In some embodiments, the first machine learning model and the second machine learning model are the same. In some embodiments, the first machine learning model and the second machine learning model are different.

[0008] In some embodiments, the generation of the first molecular data set includes applying a classifier based on an aligner to the collected metagenomic data for the first source, and the generation of the second molecular data set includes applying a classifier based on an aligner to the collected metagenomic data for the second source. In some embodiments, the generation of the first molecular data set includes applying a de novo assembler to the collected metagenomic data for the first source, and the generation of the second molecular data set includes applying a de novo assembler to the collected metagenomic data for the second source. In some embodiments, the generation of the first molecular data set includes applying a k-mer based classifier to the collected metagenomic data for the first source, and the generation of the second molecular data set includes applying a k-mer based classifier to the collected metagenomic data for the second source. In some embodiments, the generation of the first molecular data set includes applying a classifier to the collected metagenomic data for the first source, and the generation of the second molecular data set includes applying a classifier to the collected metagenomic data for the second source. In some embodiments, the classifier includes a category classifier. In some embodiments, the classifier includes a k-mer based classifier. In some embodiments, the first source includes a first database. In some embodiments, the second source includes a second database.

[0009] In some embodiments, the first molecular data set and the second molecular data set include polypeptides. In some embodiments, the first molecular data set and the second molecular data set include polynucleotides. In some embodiments, the first source includes a selected set of polynucleotides. In some embodiments, the selected set of polynucleotides includes one or more genomes. In some embodiments, the polynucleotides of the second molecular data set include publicly available polynucleotides. In some embodiments, the publicly available polynucleotides include one or more publicly available genomes. In some embodiments, the first source includes a selected set of polypeptides. In some embodiments, the selected set of polypeptides includes one or more proteomes. In some embodiments, the second source includes publicly available polypeptides. In some embodiments, the publicly available polypeptides include one or more publicly available proteomes.

[0010] In some embodiments, the first molecular data set and the second molecular data set include multiple taxids (taxonomic identification numbers). In some embodiments, the method also includes aggregating the first molecular data set and the second molecular data set for each taxid in the taxid. In some embodiments, the k-mer based classifier includes a taxonomer. In some embodiments, the k-mer based classifier includes KRAKEN.

[0011] In some embodiments, the method also includes the presence of one or more biological components in the biological components obtained from the sample based on the probability value from the output of the machine learning model using the training set. In some embodiments, the method also includes the absence of one or more biological components in the biological components obtained from the sample from the output of the machine learning model using the training set. In some embodiments, the sample is derived from one or more environmental sources, one or more industrial sources, one or more subjects, one or more microbial populations, or a combination thereof. In some embodiments, the polynucleotides obtained from the sample include one or more polynucleotides from one or more pathogens. In some embodiments, the generation of the first molecular data set and the second molecular data set occurs in parallel. In some embodiments, the method also includes iterating the first molecular data set. In some embodiments, the method also includes iterating the second molecular data set.

[0012] In some embodiments, a system for detecting biological components in a sample is provided, the system comprising one or more processors programmed to perform a method comprising obtaining metagenomic data, wherein the metagenomic data is obtained from biological components in the sample; generating a first molecular dataset; generating a second molecular dataset; creating a training set comprising an aggregated set of the first molecular dataset and the second molecular dataset; and training a machine learning model using the training set.

[0013] In some embodiments, the machine learning model comprises a random forest model. In some embodiments, the machine learning model comprises a deep neural network (DNN). In some embodiments, the machine learning model comprises a convolutional neural network (CNN). In some embodiments, the machine learning model comprises a support vector machine (SVM). In some embodiments, the machine learning model comprises a category classifier.

[0014] In some embodiments, the one or more processors are further programmed to perform a method comprising selecting a first machine learning model based on one or more metrics of the first molecular data set. In some embodiments, the one or more processors are further programmed to perform a method comprising selecting a second machine learning model based on one or more metrics of the second molecular data set. In some embodiments, the first machine learning model and the second machine learning model are the same. In some embodiments, the first machine learning model and the second machine learning model are different.

[0015] In some embodiments, the generation of the first molecular data set includes applying a classifier based on an aligner to the collected metagenomic data for the first source, and the generation of the second molecular data set includes applying a classifier based on an aligner to the collected metagenomic data for the second source. In some embodiments, the generation of the first molecular data set includes applying a de novo assembler to the collected metagenomic data for the first source, and the generation of the second molecular data set includes applying a de novo assembler to the collected metagenomic data for the second source. In some embodiments, the generation of the first molecular data set includes applying a k-mer-based classifier to the collected metagenomic data for the first source, and the generation of the second molecular data set includes applying a k-mer-based classifier to the collected metagenomic data for the second source. In some embodiments, the generation of the first molecular data set includes applying a classifier to the collected metagenomic data for the first source, and the generation of the second molecular data set includes applying a classifier to the collected metagenomic data for the second source. In some embodiments, the classifier includes a category classifier. In some embodiments, the classifier includes a k-mer-based classifier. In some embodiments, the first source includes a first database.

[0016] In some embodiments, the second source comprises a second database. In some embodiments, the first molecular data set and the second molecular data set comprise polypeptides. In some embodiments, the first molecular data set and the second molecular data set comprise polynucleotides. In some embodiments, the first source comprises a selected set of polynucleotides. In some embodiments, the selected set of polynucleotides comprises one or more genomes. In some embodiments, the polynucleotides of the second molecular data set comprise publicly available polynucleotides. In some embodiments, the publicly available polynucleotides comprise one or more publicly available genomes. In some embodiments, the first source comprises a selected set of polypeptides. In some embodiments, the selected set of polypeptides comprises one or more proteomes. In some embodiments, the second source comprises publicly available polypeptides. In some embodiments, the publicly available polypeptides comprise one or more publicly available proteomes. In some embodiments, the first molecular data set and the second molecular data set comprise a plurality of taxids (taxonomic identification numbers). In some embodiments, one or more processors are further programmed to execute a method comprising aggregating the first molecular data set and the second molecular data set for each taxid in the taxid. In some embodiments, the k-mer-based classifier comprises a taxonomer. In some embodiments, the k-mer-based classifier comprises KRAKEN.

[0017] In some embodiments, one or more processors are also programmed to perform a method including detecting the presence of one or more biological components in a biological component obtained from a sample based on a probability value from the output of a machine learning model using a training set. In some embodiments, one or more processors are also programmed to perform a method including detecting the absence of one or more biological components in a biological component obtained from a sample from the output of a machine learning model using a training set. In some embodiments, the sample is derived from one or more environmental sources, one or more industrial sources, one or more subjects, one or more microbial populations, or a combination thereof. In some embodiments, the polynucleotides obtained from the sample include one or more polynucleotides from one or more pathogens. In some embodiments, the generation of the first molecular data set and the second molecular data set occurs in parallel. In some embodiments, one or more processors are also programmed to perform a method including iterating the generation of the first molecular data set. In some embodiments, one or more processors are also programmed to perform a method including iterating the generation of the second molecular data set. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The novel features of the present invention are particularly set forth in the appended claims. The features and advantages of the present invention may be better understood by reference to the following detailed description, which sets forth illustrative embodiments utilizing the principles of the present invention, and the accompanying drawings (also referred to herein as "figures"), in which:

[0019] Figure 1A to Figure 1B Standard metagenomics pipeline metrics for 2 different pathogens are shown. Figure 1A Coverage data for the K. pneumoniae genome in samples where K. pneumoniae is expected (1) and where K. pneumoniae is expected not (0) are shown. Figure 1B Shown are read count data for the genome of E. coli in samples where E. coli is expected to be present (1) and where E. coli is expected to be absent (0).

[0020] Figure 2A is a flowchart illustrating a workflow for training a machine learning (ML) model according to some embodiments. Figure 2B is a flowchart illustrating a workflow for training a machine learning (ML) model according to some embodiments.

[0021] Figure 3A is a flow chart illustrating an iterative workflow for collecting metagenomic data from a sample and training an ML model according to some embodiments. Figure 3B is a flow chart illustrating an iterative workflow for collecting metagenomic data from a sample and training an ML model according to some embodiments.

[0022] Figure 4Ais a flow chart illustrating the workflow for collecting metagenomic data from a sample and training an ML model according to some embodiments. Figure 4B is a flow chart illustrating the workflow for collecting metagenomic data from a sample and training an ML model according to some embodiments.

[0023] Figure 5 is a flowchart illustrating the workflow for training an ML model according to some embodiments.

[0024] Figure 6 The production metagenomics pipeline (x-axis) is shown with Figure 5 Comparison of F-β scores between the metagenomics pipelines (y-axis) of trained ML models. DETAILED DESCRIPTION

[0025] Although various embodiments of the present invention have been shown and described herein, it will be apparent to those skilled in the art that these embodiments are provided by way of example only. Without departing from the present invention, those skilled in the art may conceive of many variations, changes and replacements. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed.

[0026] Accurately detecting biological components from samples from a variety of different sources is challenging. Depending on the source of the sample, a metagenomics pipeline can generate dozens of identification results from a single biological sample. The huge amount of detection may be overwhelming and difficult to interpret. In addition, it may be difficult to know which detected biological components are true positives and which are false positives. An incomplete list of reasons for this uncertainty includes reference sequence annotation errors, high genetic similarity between biological references, incomplete reference databases (e.g., query sequences are not directly represented), and sample artifacts. Standard metagenomics pipelines typically rely on individual metrics such as genome coverage or read counts to filter detection results. Using individual metrics with set thresholds is generally not enough to reliably detect many important biological components. Even if multiple metrics are used together with set thresholds, normal sample variability (e.g., changing organism composition; changing sample types, etc.) may not be accommodated. Therefore, metagenomic data analysis with the state of the art using known pipelines lacks specificity and sensitivity (unsatisfactory levels of false positives and false negatives), lacks scalability, and is usually time-consuming. In order to overcome the shortcomings of the current threshold-based methods, various embodiments of the system and method for training models to utilize machine learning are provided herein to better distinguish signals and noise in metagenomic samples (including but not limited to targeted and target-independent metagenomics), thereby improving the accuracy of biological component detection / identification. With the help of this partially or completely automated method, the system and method can be extended to process thousands of samples per day, while trained technicians in the field can only process 20 to 50 samples per day. In a non-limiting example, the system and method disclosed herein are intended to use polypeptides or nucleic acids to detect elements in a molecular manner, wherein background noise competes with the signal of interest. Non-limiting examples include "shotgun" metagenomics, enrichment sequencing, and amplicon sequencing. In any one of the embodiments described herein, it is conceivable that other types of pipelines (e.g., macrotranscriptomics, macroproteomics, metabolomics, etc.) replace the metagenomics pipeline. In some embodiments, one or more other types of pipelines can be combined with one or more metagenomics pipelines in the metagenomics pipeline in the multi-omics / macro-omics method.

[0027] As used herein, the term "biological component" refers to any molecule or cell component, part, substance or entity present in, produced by or constituting an organism. Such components may include, but are not limited to, cells, cell substructures (e.g., organelles, etc.), polypeptides, proteins, enzymes, nucleic acids / polynucleotides (e.g., DNA and RNA), genes, lipids, carbohydrates, hormones, neurotransmitters, metabolites and other bioactive agents or structural molecules. Specifically, "biological components" can be separated or derived from an organism and are present in tissues, organs, blood or other body fluids. It should be understood that "biological components" can appear in their natural state, can be synthetically derived, or engineered by genetic, biochemical or other methods, but still retain their basic functional or structural characteristics. The term "biological component" also includes those components present in or constituting microorganisms such as bacteria, viruses, fungi and protozoa. These components may include bacterial cell walls, viral capsids, fungal spores, protozoan cysts or genetic material contained in these entities. The term "biological component" also includes components that form a part of the virulence characteristics of an organism or antimicrobial resistance, such as virulence factors and antimicrobial resistance markers. As used herein, "virulence factors" include, but are not limited to, proteins, toxins, or other molecules that enhance the ability of pathogens to infect and cause disease in host organisms. As used herein, "antimicrobial resistance markers" (AMR) refer to genes, proteins, or other molecular structures or mechanisms that confer resistance to antimicrobial agents such as antibiotics, antifungals, or antivirals. In addition, "biological components" may refer to those molecules or structures or their derivatives that form part of the response of an organism to internal or external stimuli. These include, but are not limited to, antibodies, antigens, signaling molecules, and other components of the immune, endocrine, or nervous systems. The term is not limited to living or active components, and may also refer to components that have been inactivated, degraded, or otherwise modified but retain some relevant biological, structural, or functional properties. In some embodiments, the detection and identification of biological components include confirming the presence of biological entities (taxonomic groups) or characterizing genomic markers of specific phenotypes, such as drug resistance, pathogenicity, specific strains / variants / genotypes, and any combination thereof.

[0028] As used herein, the term "pathogen" refers to any biological entity or component that can cause disease, illness or abnormality in susceptible host organisms as previously defined. The term "pathogen" includes organisms such as bacteria, viruses, fungi and protozoa, as well as prions and other entities that can cause symptomatic infection or asymptomatic infection when contacting susceptible hosts. Pathogens may have and express one or more virulence factors, including but not limited to toxins, surface proteins, enzymes and other molecules, which can enhance their ability to infect and cause disease in host organisms. In addition, pathogens may carry antimicrobial resistance markers, making them tolerant to one or more antimicrobial agents, making it difficult for infection to be treated with standard therapy. Pathogens may be naturally occurring, artificially produced or genetically engineered, and may interact with the host's immune system to cause an immune response. They may exist in various forms, including but not limited to spores, cysts, free-living, intracellular or extracellular forms. They may be present in various environments, such as in water, soil, air and living organisms, and may be transmitted by a variety of pathways, including but not limited to airborne, direct contact, indirect contact, vector-borne, food-borne and waterborne pathways. The term "pathogen" also includes those entities that do not normally cause disease in a healthy host organism, but can become pathogenic under specific circumstances, such as in a host with a compromised immune system, a phenomenon known as opportunistic infection.

[0029] refer to Figure 1A to Figure 1B , Klebsiella pneumoniae and Escherichia coli are pathogens that have a great impact on human health. Figure 1A The coverage of the Klebsiella pneumoniae genome in a dataset where Klebsiella pneumoniae is expected and a dataset where Klebsiella pneumoniae is not expected is shown. As shown, there is overlap between the expected and unexpected groups, so it is not possible to use a coverage threshold to accurately distinguish true coverage signals from false coverage signals. Figure 1B The coverage of the E. coli genome in the dataset where E. coli is expected and in the dataset where E. coli is not expected is shown. Figure 1A different, Figure 1B The metric in is the count of reads from the E. coli genome that were classified into the dataset where E. coli was expected and the dataset where E. coli was not expected. However, Figure 1B The results shown in indicate the degree of overlap between the classified read counts, which makes it impossible to decide on a single read count threshold that will accurately determine when E. coli is present.

[0030] For many organisms important to human health, a single threshold is not enough to accurately determine whether an organism of interest is present in a sample. Standard metagenomics pipelines attempt to deploy a single threshold that compromises sensitivity / specificity. Additional efforts involve creating an analysis mode consisting of multiple metrics and thresholds based on a reference database and read content. Examples include the number of reads per kilobase per million mapped reads (RPKM) calculated in the Explify pipeline (Almas S et al. Deciphering Microbiota of Acute Upper Respiratory Infections: A Comparative Analysis of PCR and mNGS Methods for Lower Respiratory Trafficking Potential. Adv. Respir. Med., 91, 49-65 (2023)) and the Braken statistic calculated in the KRAKEN metagenomics pipeline (Lu J et al., Bracken: estimating species abundance in metagenomics data. Peer J Comput. Sci. 3, e104 (2017)). While these models improve upon simple genome coverage or read count thresholds, they still under-account for the amount and diversity of noise in metagenomic data.

[0031] As used herein, "k-mer" refers to a subsequence of a given length k that constitutes a sequencing read. For example, the sequence "AGCTCT" can be divided into 3-nt subsequences "AGC", "GCT", "CTC", and "TCT". In this example, each of these subsequences is a k-mer, where k = 3. K-mers can be overlapping or non-overlapping.

[0032] The sequence comparison may include one or more comparison steps in which one or more k-mers of the sequencing reads are compared to k-mers of one or more reference sequences (also referred to as "references" for short). In some embodiments, the length of the k-mer is about or greater than about 3 nt, 4 nt, 5 nt, 6 nt, 7 nt, 8 nt, 9 nt, 10 nt, 11 nt, 12 nt, 13 nt, 14 nt, 15 nt, 16 nt, 17 nt, 18 nt, 19 nt, 20 nt, 25 nt, 30 nt, 35 nt, 40 nt, 45 nt, 50 nt, 75 nt, 100 nt or longer. In some embodiments, the length of the k-mer is about or less than about 30 nt, 25 nt, 20 nt, 15 nt, 10 nt or shorter. The length of the k-mer can be in the range of 3 nt to 13 nt, 5 nt to 25 nt, 7 nt to 99 nt or 3 nt to 99 nt. The length of the k-mer analyzed in each step can be different. For example, the first comparison can compare the k-mer in the sequencing read and the reference sequence with a length of 21 nt, while the second comparison can compare the k-mer in the sequencing read and the reference sequence with a length of 7 nt. For any given sequence in the comparison step, the k-mer analyzed can be overlapping (such as in a sliding window) and can have the same or different lengths. Although k-mer is generally referred to as a nucleic acid sequence in this article, sequence comparison also includes the comparison of polypeptide sequences, including the comparison of k-mers composed of amino acids.

[0033] Machine learning provides a flexible framework for pattern learning in complex data to build predictive models that can take into account multiple covariates, including models that produce binary predictions to model the presence or absence of an organism in a sample. Several models were trained with metagenomic data to evaluate their relative performance with various molecular datasets. These models include perceptrons, logistic regression, random forests (RF), deep neural networks (DNNs), and convolutional neural networks (CNNs). Based on existing evidence, machine learning using RF models has strong generalization capabilities and can provide accurate predictions when determining whether an organism is present in a sample.

[0034] Figure 2AAn embodiment of a computer-implemented method 200 for training a machine learning model to predict the identity of a biological component in a sample is shown. The method 200 begins at a start block and then includes collecting metagenomic data from the biological component in the sample, as shown in block 202. The method also includes generating a first molecular data set, as shown in block 204, and generating a second molecular data set, as shown in block 206. In some embodiments, the number of molecular data sets is greater than two. Once the molecular data sets are created, the method 200 also includes creating a training set from the aggregation of the first molecular data set and the second molecular data set, as shown in block 208.

[0035] In some embodiments, the number of molecular data sets is 3, 4, 5, 6, 7, 8, 9, 10, 50, 100, 200, 500, 1000, 2500, 5000, 10000, 50000, 100000, 500000, 1000000 or more, or a numerical value within a range defined by any two of the foregoing values. Once the training set is created, the method 200 also includes training the machine learning model using the training set, as shown in box 210. Based on the output provided by the trained ML model, a prediction of the identity of the biological component from the sample is performed at box 212. In some embodiments, based on the output provided by the trained ML model, a prediction of the presence of a specific biological component present in the sample is provided. In some embodiments, based on the output provided by the trained ML model, a prediction of the absence of a specific biological component from the sample is provided. In some embodiments, a first molecular data set and a second molecular data set are iteratively generated, which further updates the training set, thereby continuing to train the ML model.

[0036] Figure 2B Another embodiment of a computer-implemented method 250 for training a machine learning model to predict the identity of a biological component in a sample is shown. The method 250 begins at a start block and then includes collecting metagenomic data from the biological component in the sample, as shown in block 252. The method also includes generating a molecular dataset, as shown in block 254. Once the molecular dataset is created, the method 250 also includes creating a training set from the molecular dataset, as shown in block 258.

[0037] Once the training set is created, method 250 also includes training the machine learning model using the training set, as shown in block 260. Based on the output provided by the trained ML model, a prediction of the identity of the biological component from the sample is made at block 262. In some embodiments, a prediction of the presence of a particular biological component present in the sample is provided based on the output provided by the trained ML model. In some embodiments, a prediction of the absence of a particular biological component from the sample is provided based on the output provided by the trained ML model. In some embodiments, a molecular data set is iteratively generated, which further updates the training set, thereby continuing to train the ML model.

[0038] Figure 3A An alternative computer-implemented method 300 for training a machine learning model for detecting biological components in a sample is shown. The method 300 begins at a start state, and then the method 300 includes collecting metagenomic data from biological components in the sample, as shown in box 302. Then, the method 300 applies a metagenomic classifier to a first source at box 304, and then generates a first molecular data set, as shown in box 308. The method 300 also applies a metagenomic classifier to a second source at box 306, and generates a second molecular data set, as shown in box 310. Once the first molecular data set is created at box 308, the method moves to a decision state 311A ​​to determine whether the method should be iterated. If the method should be iterated, the method 300 returns to box 304. If the method should not be iterated, the method 300 moves to box 309 to create a training set from the aggregation of molecular data sets. Similarly, once the second molecular data set is created at box 310, the method moves to a decision state 311B to determine whether the method should be iterated. If the method should be iterated, the method 300 returns to box 306. If the method should not iterate, the method 300 moves to block 309 to create a training set from an aggregation of the molecular data sets.

[0039] In some embodiments, the number of molecular data sets is greater than two. Once the training set is created, method 300 is transferred to box 312, where the ML model can be trained. In some embodiments, the number of molecular data sets is 3, 4, 5, 6, 7, 8, 9, 10, 50, 100, 200, 500, 1000, 2500, 5000, 10000, 50000, 100000, 500000, 1000000 or more, or a numerical value within the range defined by any two of the aforementioned values. Based on the output provided by the trained ML model, a prediction of the identity of the biological component from the sample is performed at box 314. In some embodiments, based on the output provided by the trained ML model, a prediction of the presence of the biological component from the sample is provided. In some embodiments, based on the output provided by the trained ML model, a prediction of the absence of the biological component from the sample is provided.

[0040] Figure 3B Another alternative computer-implemented method 350 for training a machine learning model for detecting biological components in a sample is shown. The method 350 begins at a start state, and then the method 350 includes collecting metagenomic data from biological components in the sample, as shown in box 352. Then, the method 350 applies a metagenomic classifier to the source at box 354, and then generates a first molecular data set, as shown in box 358. Once the molecular data set is created at box 358, the method moves to a decision state 361 to determine whether the method should be iterated. If the method should be iterated, the method 350 returns to box 354. If the method should not be iterated, the method 350 moves to box 359 to create a training set from the molecular data set.

[0041] Once one or more training sets are created, method 350 moves to block 362 where an ML model can be trained. Based on the output provided by the trained ML model, a prediction of the identity of a biological component from the sample is made at block 364. In some embodiments, a prediction of the presence of a biological component from the sample is provided based on the output provided by the trained ML model. In some embodiments, a prediction of the absence of a biological component from the sample is provided based on the output provided by the trained ML model.

[0042] Figure 4AA computer-implemented method 400 for training a machine learning model for detecting biological components in a sample is shown. The method includes collecting metagenomic data from biological components in the sample, as shown in block 402. The method also includes applying a metagenomics classifier to the collected biological components from the sample against a first database, as shown in block 404. The method then moves to block 408 to generate a first molecular data set. In some embodiments, the first database includes selected sequence data. In some embodiments, the selected sequence data includes polynucleotide sequences. In some embodiments, the selected sequence database includes polypeptide sequences.

[0043] After the method collects metagenomic data from the sample at box 402, the method also transfers to box 406, where the metagenomic classifier is applied to the collected biological components from the sample for the second database. The method then transfers to box 410 to generate a second molecular data set. In some embodiments, the second database includes publicly available sequence data. In some embodiments, the publicly available sequence data includes polynucleotides. In some embodiments, the publicly available sequence data includes polynucleotides. Once the first molecular data set and the second molecular data set are created, the method includes creating a training set at box 412, which includes an aggregated set of the first molecular data set and the second molecular data set. Based on the output provided by the trained ML model, the method transfers to box 414 to train the ML model. Based on the output provided by the trained ML model, the method transfers to box 416 to predict the identity of the biological component from the sample. In some embodiments, based on the output provided by the trained ML model, a prediction of the absence of the biological component from the sample is provided. In some embodiments, the first molecular data set and the second molecular data set are iteratively generated, which further updates the training set, thereby continuing to train the ML model.

[0044] In some embodiments, the machine learning model includes a random forest model. In some embodiments, the machine learning model includes a deep neural network (DNN). In some embodiments, the machine learning model includes a convolutional neural network (CNN). In some embodiments, the machine learning model includes a support vector machine. In some embodiments, the machine learning model includes a category classifier. In some embodiments, the method also includes selecting a first machine learning model based on one or more metrics of a first molecular data set. In some embodiments, the method also includes selecting a second machine learning model based on one or more metrics of a second molecular data set. In some embodiments, the metrics for the selection of the machine learning model include the abundance of data available for training, the relative difficulty of generating a model for a specific organism. In some embodiments, the type of machine learning model is the same for each molecular data set. In some embodiments, the type of machine learning model is different for each molecular data set. In some embodiments, the type of machine learning model is the same for some molecular data sets in the molecular data set, and is different for other molecular data sets in multiple molecular data sets. In some embodiments, different types of ML models can be used for different organisms.

[0045] In some embodiments, the generation of the first molecular data set includes applying a classifier based on an aligner to the collected metagenomic data for the first data source, and the generation of the second molecular data set includes applying a classifier based on an aligner to the collected metagenomic data for the second data source. In some embodiments, the generation of the first molecular data set includes applying a de novo assembler to the collected metagenomic data for the first data source, and the generation of the second molecular data set includes applying a de novo assembler to the collected metagenomic data for the second data source. In some embodiments, the generation of the first molecular data set includes applying a k-mer based classifier to the collected metagenomic data for the first data source, and the generation of the second molecular data set includes applying a k-mer based classifier to the collected metagenomic data for the second data source. In some embodiments, the generation of the first molecular data set includes applying a classifier to the collected metagenomic data for the first data source, and the generation of the second molecular data set includes applying a classifier to the collected metagenomic data for the second data source. In some embodiments, the classifier includes a category classifier. In some embodiments, the classifier includes a k-mer based classifier. In some embodiments, the classifier applied to the first molecular data set is different from the classifier applied to the second molecular data set. In some embodiments, the classifier applied to the first molecular dataset is the same as the classifier applied to the second molecular dataset.

[0046] In some embodiments, the first molecular data set and the second molecular data set include polypeptides. In some embodiments, the first molecular data set and the second molecular data set include polynucleotides. In some embodiments, the first data source includes a selected set of polynucleotides. In some embodiments, the selected set of polynucleotides includes one or more genomes. In some embodiments, the polynucleotides of the second molecular data set include publicly available polynucleotides. In some embodiments, the publicly available polynucleotides include one or more publicly available genomes. In some embodiments, the first data source includes a selected set of polypeptides. In some embodiments, the selected set of polypeptides includes one or more proteomes. In some embodiments, the second data source includes publicly available polypeptides. In some embodiments, the publicly available polypeptides include one or more publicly available proteomes.

[0047] Figure 4B Another computer-implemented method 450 of training a machine learning model for detecting biological components in a sample is shown. The method includes collecting metagenomic data from the biological components in the sample, as shown in box 452. The method also includes applying a metagenomics classifier to the collected biological components from the sample against a database, as shown in box 454. The method then moves to box 458 to generate a molecular data set. In some embodiments, the database includes selected sequence data. In some embodiments, the selected sequence data includes polynucleotide sequences. In some embodiments, the selected sequence database includes polypeptide sequences. In some embodiments, the database includes publicly available sequence data. In some embodiments, the publicly available sequence data includes polynucleotides.

[0048] In some embodiments, the publicly available sequence data includes polynucleotides. Once the first molecular dataset and the second molecular dataset are created, the method includes creating a training set at box 462. Based on the output provided by the trained ML model, the method transfers to box 464 to train the ML model. Based on the output provided by the trained ML model, the method transfers to box 466 to predict the identity of the biological component from the sample. In some embodiments, based on the output provided by the trained ML model, a prediction of the absence of the biological component from the sample is provided. In some embodiments, the molecular dataset is iteratively generated, which further updates the training set, thereby continuing to train the ML model.

[0049] In some embodiments, the machine learning model comprises a random forest model. In some embodiments, the machine learning model comprises a deep neural network (DNN). In some embodiments, the machine learning model comprises a convolutional neural network (CNN). In some embodiments, the machine learning model comprises a support vector machine. In some embodiments, the machine learning model comprises a category classifier.

[0050] In some embodiments, the generation of the molecular data set includes applying a classifier based on an aligner to the collected metagenomic data for the data source. In some embodiments, the generation of the molecular data set includes applying a de novo assembler to the collected metagenomic data for the data source. In some embodiments, the generation of the molecular data set includes applying a k-mer based classifier to the collected metagenomic data for the data source. In some embodiments, the generation of the molecular data set includes applying a classifier to the collected metagenomic data for the data source. In some embodiments, the classifier includes a category classifier. In some embodiments, the classifier includes a k-mer based classifier.

[0051] In some embodiments, the molecular data set includes polypeptides. In some embodiments, the molecular data set includes polynucleotides. In some embodiments, the data source includes a selected set of polynucleotides. In some embodiments, the selected set of polynucleotides includes one or more genomes. In some embodiments, the molecular data set includes publicly available polynucleotides. In some embodiments, the publicly available polynucleotides include one or more publicly available genomes. In some embodiments, the data source includes a selected set of polypeptides. In some embodiments, the selected set of polypeptides includes one or more proteomes. In some embodiments, the data source includes publicly available polypeptides. In some embodiments, the publicly available polypeptides include one or more publicly available proteomes.

[0052] Figure 5A computer-implemented method 500 for training a machine learning model for detecting biological components in a sample is shown. The method includes collecting metagenomic data from biological components in the sample in box 502. The method also includes: generating a first k-mer data set by applying a k-mer-based classifier to the metagenomic data for a selected set of organism genomes at box 504; generating a second k-mer data set by applying a k-mer-based classifier to the metagenomic data for a subset of nucleotides from a private or publicly available database (e.g., a subset of the NCBI blast database) at box 506; aggregating the output of the classifier from each taxonomic group identity (taxid) at box 506; and training the machine learning model using the training set at box 508. Based on the output provided by the trained ML model, the method moves to box 510 to predict the identity of the biological component from the sample. In some embodiments, the number of molecular data sets is greater than two. In some embodiments, the number of molecular data sets is 3, 4, 5, 6, 7, 8, 9, 10, 50, 100, 200, 500, 1000, 2500, 5000, 10000, 50000, 100000, 500000, 1000000 or more, or a value within a range defined by any two of the foregoing values. Figure 5 As shown, in order to construct a machine learning model for predicting the presence of a type of nucleic acid in a sample, the read data can be processed by a bioinformatics pipeline with the goal of training a machine learning model by constructing a training set. In some embodiments, the first k-mer data set and the second k-mer data set include multiple taxids. In some embodiments, the method also includes aggregating the first k-mer data set and the second k-mer data set for each taxid in the taxid. In some embodiments, the k-mer-based classifier includes or can extend Taxonomer functionality. Taxonomers are discussed in U.S. Patent No. 11,335,436, the contents of which are incorporated herein by reference in their entirety. In some embodiments, the k-mer-based classifier includes KRAKEN (genomebiology.biomedcentral.com / articles / 10.1186 / s13059-016-0969-1).

[0053] In some embodiments, the sample from which the polynucleotides for the analysis performed by the methods and systems of the present invention can be derived can be from any of a variety of sources. Non-limiting examples of sample sources include environmental sources, industrial sources, one or more subjects, and one or more microbial populations. Examples of environmental sources include, but are not limited to, farmland, soil, dust, lakes, rivers, oceans, reservoirs, vents, walls, roofs, soil samples, plants, and swimming pools. Examples of industrial sources include, but are not limited to, clean rooms, hospitals, food processing areas, food production areas, food materials, medical laboratories, pharmacies, drug compounding centers, and wastewater treatment plants. Biological components can be isolated from the cystic algae kingdom such as malaria and dinoflagellates. Non-limiting examples of subjects from which biological components can be isolated include multicellular organisms, such as fish, amphibians, reptiles, birds, and mammals. Non-limiting examples of mammals include primates (e.g., apes, monkeys, gorillas), rodents (e.g., mice, rats), farm animals (e.g., cattle, pigs, sheep, horses), dogs, cats, or rabbits. In some embodiments, the mammal is a human. In some embodiments, the mammal is a single subject. Samples may include samples from subjects, such as biological fluids; whole blood; blood products; red blood cells; white blood cells; buffy coat; swabs; urine; sputum; saliva; nasal mucus; mucus; semen; lymph; amniotic fluid; cerebrospinal fluid; peritoneal effusion; pleural effusion; biopsy specimens; fluid from cysts; synovial fluid; vitreous fluid; aqueous humor; bursal fluid; eyewash; eye aspirate; plasma; serum; lung lavage fluid; lung aspirate; animals, including humans, tissues, including but not limited to liver, spleen, kidney, lung, intestine, brain, heart, muscle, pancreas, cell culture, and lysates, extracts or substances and fractions obtained from the above samples, or any cells, microorganisms and viruses that may be present on or in the sample. Also included are tissues, cells and their progeny of biological entities obtained in vivo or cultured in vitro.

[0054] The sample may include cells of primary culture or cell line. Examples of cell lines include, but are not limited to, 293-T human kidney cells, A2870 human ovarian cells, A431 human epithelial cells, B35 rat neuroblastoma cells, BHK-21 hamster kidney cells, BR293 human breast cells, CHO Chinese hamster ovary cells, CORL23 human lung cells, HeLa cells, or Jurkat cells. The sample may include uniform or mixed microbial populations, including one or more of viruses, bacteria, protozoa, ankaryotic protozoa, vesicle algae, archaea, or fungi. In some embodiments, the microorganism is a pathogen. In some embodiments, the microorganism is a human pathogen. Examples of viruses include, but are not limited to, Ebola virus, hepatitis virus, herpes virus, human immunodeficiency virus, influenza virus, lettuce vein-associated virus, mosaic virus, rhinovirus, ringspot virus, rotavirus, West Nile virus. Examples of bacteria include, but are not limited to, Bacillus cereus, Citrobacter krusei, Clostridium perfringens, Escherichia coli, Enterobacter aerogenes, Enterococcus faecalis, Klebsiella pneumoniae, Lactobacillus acidophilus, Listeria monocytogenes, Micrococcus luteus, Propionibacterium granulosum, Pseudomonas aeruginosa, Serratia marcescens, Staphylococcus aureus, Staphylococcus aureus Mu3, Staphylococcus aureus Mu50, Staphylococcus epidermidis, Staphylococcus aegypti, Streptococcus agalactiae, Streptococcus pneumoniae, Streptococcus pyogenes, and Yersinia enterocolitica. Examples of fungi include, but are not limited to, Absidia agglomerata, Aspergillus niger, Aspergillus niger, Candida albicans, Candida albicans, Geotrichum candidum, Geotrichum candidum, Hansenula anomala, Hansenula anomala, Microsporum gypsum, Microsporum gypsum, Candida genus, Candida genus, Mucor, Mucor, Penicillium expansum, Penicillium corymbiferum, Rhizopus, Rhodotorula, Pasteurella, Saccharomyces cerevisiae, and Saccharomyces cerevisiae. The sample may also be a processed sample, such as a preserved, fixed, and / or stabilized sample. The sample may comprise or consist essentially of a polypeptide. The sample may comprise or consist essentially of a polynucleotide. The sample may comprise or consist essentially of an RNA. The sample may comprise or consist essentially of a DNA. In some embodiments, cell-free polynucleotides (e.g., cell-free DNA and / or cell-free RNA) are analyzed. Generally, cell-free polynucleotides are extracellular polynucleotides present in a sample (e.g., a sample from which cells have been removed, a sample that has not been subjected to a lysis step, or a sample that has been treated to separate cellular polynucleotides from extracellular polynucleotides). In one non-limiting example, cell-free polynucleotides include polynucleotides that are released into the circulation upon cell death and are isolated as cell-free polynucleotides from the plasma fraction of a blood sample.

[0055] Methods for extracting and purifying nucleic acids are well known in the art. For example, nucleic acids can be purified by organic extraction with phenol, phenol / chloroform / isoamyl alcohol, or similar formulations (including TRIzol and TriReagent). Other non-limiting examples of extraction techniques include: (1) organic extraction followed by ethanol precipitation, for example, using phenol / chloroform organic reagents, with or without an automated nucleic acid extractor, for example, a Model 341 DNA Extractor from Applied Biosystems (Foster City, Calif.); (2) stationary phase adsorption methods; and (3) salt-induced precipitation methods, which are often referred to as "salting-out" methods. Another example of nucleic acid separation and / or purification includes the use of magnetic particles to which nucleic acids can bind specifically or non-specifically, followed by separation of beads and washing with a magnet, and elution of nucleic acids from the beads. In some embodiments, an enzymatic digestion step may be performed prior to the above separation method to help eliminate unwanted proteins from the sample, such as digestion with proteinase K and / or other similar proteases. If desired, an RNase inhibitor may be added to the lysis buffer. For certain cell or sample types, it may be necessary to add a protein denaturation / digestion step to the protocol. The purification method may involve separation of DNA, RNA, or both. When both DNA and RNA are separated together during or after the extraction process, further steps may be used to purify one or both of them. The extracted nucleic acid may also produce subfractions, such as purified by size, sequence, or other physical or chemical methods.

[0056] In some embodiments, the sample obtained from one or more sources can be directly used. In some embodiments, the sample can be pre-treated to change the characteristics of the sample. In some embodiments, such pre-treatment may include preparing plasma, diluting viscous fluids, etc. from blood. In some embodiments, the method of pre-treatment may also include, but is not limited to, filtering, precipitation, dilution, distillation, mixing, centrifugation, freezing, lyophilization, concentration, amplification, nucleic acid fragmentation, interfering component inactivation, adding reagents, cracking, etc. If such pre-treatment methods are adopted for sample, such pre-treatment methods usually make the nucleic acid of interest remain in the test sample, and sometimes its concentration is proportional to the concentration in the untreated test sample (for example, a sample that is not subjected to any such pre-treatment methods).

[0057] In some embodiments, each sample structured data file is mapped to two or more separate databases. In some embodiments, the structured data file includes a FASTQ file. In some embodiments, the first database includes a first group of polynucleotides. In some embodiments, the second database includes a second group of polynucleotides. In some embodiments, the first group of polynucleotides includes a selected set of polynucleotides. In some embodiments, the selected set of polynucleotides includes one or more genomes. In some embodiments, the selected set of polynucleotides includes one or more nucleotide sequences. In some embodiments, the selected set of nucleotide sequences corresponds to one or more taxa. In some embodiments, based on the sequences analyzed by any method described herein, at least 1, 2, 5, 10, 25, 50, 50, 100, 250, 500, 1000, 5000, 10000, 50000, 100000, 250000, 500000, 1000000 or more or different taxa in the range defined by any two of the aforementioned values ​​are identified as absent or present (and optionally abundance, which can be relative). In some embodiments, the second group of polynucleotides includes publicly available polynucleotides. In some embodiments, publicly available polynucleotides include one or more publicly available genomes. Exemplary databases providing publicly available polynucleotides include the Sequence Read Archive (SRA) database, which is the largest publicly available public repository of high-throughput sequencing data. The National Center for Biotechnology Information (NCBI) hosts the SRA database, which includes more than 5100 trillion total base pairs.

[0058] In some embodiments, the generation of the first data set and the second data set occurs in parallel. In some embodiments, the generation of the data set occurs successively. In some embodiments, the generation of the data set occurs according to a schedule. In some embodiments, more than two data sets will be generated. In some embodiments, the method also includes detecting a type of polynucleotide obtained from a sample from the output of a machine learning model using a training set. In some embodiments, the machine learning model includes a random forest (RF) model. In some embodiments, the machine learning model includes a deep neural network (DNN). In some embodiments, the machine learning model includes a support vector machine (SVM). In some embodiments, the machine learning model includes a convolutional neural network (CNN). In some embodiments, the machine learning model includes a category classifier.

[0059] In some embodiments, molecular attributes or covariates are extracted from each molecular data set and placed in a table, where the covariates can be used for metagenomics model training / testing. Metagenomics k-mer-based classifiers can use HyperLogLog (en.wikipedia.org / wiki / HyperLogLog) data structures to track the number of different k-mers from reads assigned to taxid. The data structure provides how many different data (as opposed to a single repeat sequence) are assigned to taxid. When calculating the reference index (before any classification), the metagenomics classifier can also calculate the value of each taxid based on the reference sequence. With this data, the percentage of different k-mers that appear in the sample compared to the reference database can be ascertained.

[0060] In some embodiments, metagenomic data covariates are collected at the taxid level. In some embodiments, covariates include read counts (from the first database and the second database), the intersection of read identifiers classified from different databases to the same taxid, different k-mer counts, median classification read scores, or a combination thereof. In a machine learning model using an RF model, covariate importance is calculated when the RF model is established. A non-limiting example of covariate importance is provided in Table 1 below.

[0061] Table 1. Non-limiting example of feature importances assigned when building RF models using XGBoost

[0062] Covariates importance Database 1 read count 0.03623812 Database 2 read count 0.01373014 Read counts of two databases 0.01377384 Distinct k-mer counts 0.83370483 Median classified read score 0.09675267 Median NT score 0.00580038

[0063] In some embodiments, the method further comprises detecting the presence of biological components obtained from the sample from the output of the machine learning model using the training set. In some embodiments, the presence of biological components is provided in the form of probability values ​​from the output of the machine learning trained on the training set. In some embodiments, the probability value is taken as a face value, so that the probability value of the predicted / identified biological component higher than 0.5 indicates the presence. In some embodiments, the probability value is greater than 0.5 according to the biological component. In some embodiments, the probability value is adjusted based on the training set. In some embodiments, the training set is improved on the basis of the older training set. In some embodiments, the method further comprises detecting the absence of biological components obtained from the sample from the output of the machine learning model using the training set. In some embodiments, detection also comprises detecting the absence of biological components obtained from the sample from the output of the machine learning model using the training set. In some embodiments, the sample is derived from one or more environmental sources, one or more industrial sources, one or more subjects, one or more microbial populations, or a combination thereof. In some embodiments, the biological components obtained from the sample include biological components from one or more pathogens.

[0064] In some embodiments, a system for detecting biological components in a sample is provided, the system comprising a computer-readable medium storing instructions and one or more processors programmed to read the instructions to perform a method of any of the embodiments provided herein.

[0065] In implementation, given the available data in the training set and the model being appropriately generalized, specific model selection is not as important as being able to control overfitting. It is expected that the RF model in the XGBoost framework provides enough parameters to tune the model to prevent overfitting, even if the amount of data in the training set is relatively small. Another desired attribute of using the RF model from XGBoost is the ability to perform online learning. As more data is processed, with appropriate markings, the machine learning model can be updated "online" so that the machine learning model does not have to be rebuilt whenever the machine learning model is trained with new data. As used herein, "online learning", which is interchangeable with "incremental learning" and "out-of-core learning", is a machine learning paradigm in which the model incrementally updates its knowledge as new data instances become available, rather than processing the entire data set in batches. This method is particularly useful when processing large-scale data sets that cannot fit into memory or when receiving data in a continuous stream.

[0066] Example 1

[0067] To establish a training set for the machine learning model, multiple sequence datasets known to have signals of microorganisms from multiple sources are used for the training set, including public database sources, environmental sources, industrial sources, subjects, or a combination thereof. The machine learning model is trained based on such training sets of known microorganisms. The method is iterated for multiple known microorganisms. After training the machine learning model from multiple training sets, such as Figure 5 As shown, a sample that may include microorganisms of unknown identity is processed, whereby the sample is processed and nucleic acids (if any) in the sample are sequenced to obtain sequence data.

[0068] For this example, the sequence data is stored in FASTQ format. Alternative formats may be used. The sequence data is then transformed by a k-mer based classifier and / or aligner and / or assembler to generate labeled data about microorganisms that may be present in the sample. The labeled data forms at least a portion of a training set for training a machine learning model. When the machine learning model is trained on a training set including a data set of various human pathogens, subsequent predictions of human pathogens in the sample using the trained model show improved accuracy and sensitivity compared to metagenomics pipelines on the market that use various standard metrics and set thresholds.

[0069] A single model for each organism is usually not effective. It is effective to build multiple models for each organism, because different organisms have different genomic landscapes, they also exist in different communities and may have more closely related neighbors than, for example, a smaller number of microorganisms. Having more than one model per organism involves many genomic factors.

[0070] The number of different data sets for a particular organism required to obtain a training set for training a machine learning model to make accurate predictions varies from organism to organism depending on the genomes being compared. Generally speaking, at least 20 or more positive samples of a particular organism are sufficient for a machine learning model to produce an accurate prediction of a microorganism (e.g., the identity of the microorganism). However, some pathogens such as E. coli require at least 100 or more positive samples to effectively distinguish signal and noise from other microorganisms that may be present in the sample to be tested. This higher count is due to the possibility of many species of nucleic acids being present in a sample and how E. coli resembles other microorganisms of the same family that coexist in a sample, so it is important to distinguish these differences.

[0071] Example 2

[0072] In order to build machine learning models to aid in the detection of organisms, multiple sequencing datasets can be used. These datasets generally fall into one of two categories: 1. Simulated datasets, where reads are generated in silico and the content is completely known and controlled; 2. Real sequencing experiments derived from known samples or from samples whose sequencing content has been characterized. Sequencing content from real sequencing experiments often contains unexpected reads due to reasons such as contamination of sequencing reagents or other unexpected organisms present in the sample environment.

[0073] Sequencing datasets from real samples are often more effective for training machine learning models than their simulated counterparts for detecting microorganisms because the data contain noise and organismal genomic variation present in real samples. However, for practical reasons, it may be difficult to build a training set consisting only of real sequencing experiments. In addition, simulated datasets are often required to augment the training set in order to provide a complete and robust ML model.

[0074] Multiple sequencing data sets are typically used for each organism to be detected by the model. Generally speaking, at least 20 positive data sets for each organism (usually a mixture of real samples and simulated samples) are used to effectively train the model; more training data generally leads to better models. Some organisms require more training data than other organisms to effectively distinguish signals from noise. For example, organism A may need 200 samples to produce accurate predictions, while another organism may only need 10 samples because the organisms have different genomic situations, exist in different communities, and have different levels of genetic correlation with other organisms. For these reasons, a single model applied to all organisms is usually not as effective as establishing a separate model for each organism.

[0075] The training dataset is processed by sending each sample through a k-mer classifier that will classify it against a set of reference sequences. Figure 5 In this case, this may mean that each sample in the training dataset will be classified at least twice, once for each set of reference sequences. For each sample, the classifier outputs from two different classifications (one from each database or reference set) are combined to create an aggregate covariate for training the model. The organism model is then trained using a combination of data classified against the different reference sets.

[0076] Example 3

[0077] Using genome simulations and read data from the Sequence Read Archive (SRA), we used Figure 5 The RF model in the pipeline shown generates test data to train and test the model for about 100 or more different bacterial urinary pathogens. For comparison, the same test data was processed using the Explify Urinary Pathogen ID / AMR Panel (UPIP) Data Analysis application available in Illumina BaseSpace Sequence Hub (BSSH) to determine which urinary pathogens were present in each sample. Figure 6 Results of the RF model built using the XGBoost framework for calling uropathogens in a test set of genomic simulations and SRA data were compared to the results of the Explify UPIP pipeline. Using the Fβ score (the harmonic mean between recall and precision), the pipelines using machine learning based on the RF model (y-axis) generally outperformed the Explify UPIP pipeline (x-axis).

[0078] Example 4

[0079] As described above, machine learning models can be used to significantly improve the accuracy of organism detection using k-mer based classifiers. To demonstrate this, we performed a simulation study using a k-mer classifier using simulated reads and a collection of 27,155 bacterial 16S rRNA sequences collected from NCBI in April 2024. Since the conserved nature of the 16S gene means that there are relatively small genetic differences between the 16S genes of different species, read classification and organism detection using a k-mer classifier with this set of 16S references is challenging.

[0080] The simulation study was set up as follows: 1000 different taxa represented in the collected 16S sequences were randomly selected. Then, for each of these taxa, one of the corresponding 16S sequences was selected as a simulation reference. For taxa containing more than one 16S sequence, simulation candidates were removed from the larger set to challenge the classifier with sequences that were not directly represented in the database. For taxa containing only a single reference 16S sequence, it was left in the reference set to ensure that the taxa were represented during classification. After selecting these 1000 references for simulation and updating the reference set, two different tools were built ( k-mer classifier and For each selected reference for simulation, the NCBI ART simulator was used to simulate the Figure 4B Illumina Box 452 The error pattern simulates paired-end 150bp reads at 20X read depth. Figure 4B Box 454, use k-mer classifier or The classifier classifies the reads of each of the 1000 simulated references. Figure 4B In block 458 , the classification results are aggregated into a table so that recall, precision, and accuracy can be generated.

[0081] according to Figure 4B In block 462, once the simulated data is assembled, six of the features generated by the DRAGEN k-mer classifier in the columns of the table are used to build a machine learning model using the XGBoost framework. The six features are duplicity, distinct_coverage, read_count, total_kmer_count, distinct_kmer_count, and taxid_distinct_kmer_count. For the expected result features, the information about which reference was simulated is used to label each row of the table, with the taxa expected to be present being labeled with a "1" and otherwise labeled with a "0". Figure 4B In block 464, a classification model is trained using the XGBoost framework with random subsets of the aggregated simulated data, and the trained classification model is tested using independent random subsets of the same simulated data to demonstrate the usefulness of machine learning using the above features to improve the accuracy of taxonomic groups of organisms. Table 2 shows the results of this study. As shown in Table 2, compared to using Classifier or Compared with classifiers, combined machine learning and The classifier significantly improved the performance.

[0082] Table 2. Results of the simulation study described in Example 4

[0083] tool Recall Accuracy F1-score DRAGEN .75 .66 .58 KRAKEN2 .74 .65 .57 DRAGEN+ML .93 .93 .93

[0084] definition

[0085] The section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described.

[0086] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. The use of the term "include" as well as other forms such as "include / includes / included" is not restrictive. The use of the term "have" as well as other forms such as "have / has / had" is not restrictive. As used in this specification, the terms "comprise" and "includes" are to be interpreted as having an open-ended meaning, whether in a transitional phrase or in the body of the claims. That is, the above terms are to be interpreted synonymously with the phrases "at least have" or "at least include". For example, when used in the context of a process, the term "comprises" means that the process includes at least the listed steps, but may also include additional steps. When used in the context of a compound, composition, or device, the term "comprises" means that the compound, composition, or device includes at least the listed features or components, but may also include additional features or components.

[0087] The terms "polynucleotide", "oligonucleotide", "nucleic acid" and "nucleic acid molecule" are used interchangeably herein and refer to a sequence of covalently linked nucleotides (i.e., ribonucleotides of RNA, deoxyribonucleotides of DNA, analogs thereof, or mixtures thereof) of any length, wherein the 3' position of the pentose of one nucleotide is connected to the 5' position of the pentose of the next nucleotide by a phosphodiester group. These terms are understood to include analogs of DNA, RNA, cDNA or antibody-oligonucleotide conjugates made of nucleotide analogs as equivalents, and are applicable to single-stranded (such as sense or antisense) and double-stranded polynucleotides. As used herein, the term also encompasses cDNA, i.e., complementary DNA or copy DNA produced by an RNA template, for example, by the action of a reverse transcriptase. The term refers only to the primary structure of a molecule. Therefore, the term includes, but is not limited to, triple-stranded, double-stranded and single-stranded deoxyribonucleic acids ("DNA"), and triple-stranded, double-stranded and single-stranded ribonucleic acids ("RNA"). Nucleotide includes a sequence of any form of nucleic acid.

[0088] Additional Notes

[0089] Various embodiments of the present disclosure may be systems, methods and / or computer program products at any possible level of technical detail integration. A computer program product may include a computer-readable storage medium (or multiple media) having computer-readable program instructions thereon for causing a processor to perform various aspects of the present disclosure.

[0090] For example, the functions described herein may be performed as software instructions executed by one or more hardware processors and / or any other suitable computing devices and / or in response to software instructions executed by one or more hardware processors and / or any other suitable computing devices. The software instructions and / or other executable code may be read from a computer-readable storage medium (or multiple media). A computer-readable storage medium may also be referred to herein as a computer-readable storage device or a computer-readable storage device.

[0091] A computer-readable storage medium may be a tangible device capable of storing and storing data and / or instructions for use by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electronic storage device (including any volatile and / or non-volatile electronic storage device), a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of computer-readable storage media includes the following: a portable computer disk, a hard disk, a solid-state drive, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device such as a punch card or a raised structure in a groove on which instructions are recorded, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium should not be interpreted as a transient signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagated by a waveguide or other transmission medium (e.g., a light pulse by a fiber optic cable), or an electrical signal transmitted by a wire.

[0092] The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to a corresponding computing / processing device, or downloaded to an external computer or external storage device via a network (e.g., the Internet, a local area network, a wide area network, and / or a wireless network). The network may include copper transmission cables, optical transmission fibers, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions to be stored in a computer-readable storage medium within the corresponding computing / processing device.

[0093] Computer-readable program instructions (also referred to herein as, for example, "code," "instructions," "modules," "applications," "software applications," etc.) for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, configuration data for an integrated circuit, or source code or object code written in any combination of one or more programming languages ​​(including object-oriented programming languages ​​such as Smalltalk, C++, etc. and procedural programming languages, such as the "C" programming language or similar programming languages). Computer-readable program instructions may be called from other instructions or from themselves, and / or may be called in response to detected events or interrupts. Computer-readable program instructions configured for execution on a computing device may be provided on a computer-readable storage medium, and / or as a digital download (and may be initially stored in a compressed or installable format that requires installation, decompression, or decryption prior to execution), which may then be stored on a computer-readable storage medium. Such computer-readable program instructions may be stored partially or completely on a memory device (e.g., a computer-readable storage medium) of an executing computing device for execution by the computing device. Computer-readable program instructions may be executed entirely on a user's computer (e.g., an execution computing device), partially on a user's computer, as a standalone software package, partially on a user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuits including, for example, programmable logic circuits, field programmable gate arrays (FPGAs), or programmable logic arrays (PLAs) may execute computer-readable program instructions to personalize the electronic circuits by utilizing the state information of the computer-readable program instructions, so as to perform various aspects of the present disclosure.

[0094] Various aspects of the present disclosure are described herein with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present disclosure. It will be understood that each block of the flowchart illustration and / or block diagram and the combination of blocks in the flowchart illustration and / or block diagram can be implemented by computer-readable program instructions.

[0095] These computer-readable program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing device create a means for implementing the functions / actions specified in one or more boxes of the flowchart and / or block diagram. These computer-readable program instructions may also be stored in a computer-readable storage medium, which may direct a computer, a programmable data processing device, and / or other equipment to operate in a particular manner, such that the computer-readable storage medium having the instructions stored therein includes an article of manufacture, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes of the flowchart and / or block diagram.

[0096] Computer readable program instructions may also be loaded onto a computer, other programmable data processing device, or other device to perform a series of operating steps on the computer, other programmable device, or other device to produce a computer-implemented process so that the instructions executed on the computer, other programmable device, or other device implement the functions / actions specified in one or more boxes of the flow chart and / or block diagram. For example, the instructions may initially be carried by a disk or solid-state drive of a remote computer. The remote computer may load instructions and / or modules into its dynamic memory and send instructions via a telephone, cable, or optical line using a modem. A modem local to the server computing system may receive data on the telephone / cable / optical line and place the data on the bus using a converter device including appropriate circuits. The bus may carry data to a memory, and the processor may retrieve and execute instructions from the memory. The instructions received by the memory may optionally be stored on a storage device (e.g., a solid-state drive) before or after being executed by a computer processor.

[0097] The flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each frame in the flow chart or block diagram may represent a part of a service, module, fragment or instruction, including one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions noted in the frame may not occur in the order noted in the figure. For example, the two frames shown in succession can actually be executed substantially simultaneously, or these frames can sometimes be executed in reverse order, depending on the functions involved. In addition, some frames can be omitted in some implementations. The methods and processes described herein are also not limited to any specific order, and the frames or states related thereto can be executed in other appropriate orders.

[0098] It should also be noted that each block in the block diagram and / or flowchart illustration, and combinations of blocks in the block diagram and / or flowchart illustration, may be implemented by a dedicated hardware-based system that performs the specified functions or actions or executes a combination of dedicated hardware and computer instructions. For example, any of the processes, methods, algorithms, elements, blocks, applications, or other functions (or portions of functions) described in the preceding sections may be embodied in electronic hardware such as a dedicated processor (e.g., an application specific integrated circuit (ASIC)), a programmable processor (e.g., a field programmable gate array (FPGA)), a dedicated circuit, and / or the like (any of which may also combine custom hardwired logic, logic circuits, ASICs, FPGAs, etc. with custom programming / execution of software instructions to implement these techniques) and / or fully or partially automated via such electronic hardware.

[0099] Any of the above processors and / or devices incorporating any of the above processors may be referred to herein as, for example, a "computer," "computer device," "computing device," "hardware computing device," "hardware processor," "processing unit," etc. The computing devices of the above embodiments may typically (but not necessarily) be controlled and / or coordinated by operating system software such as Mac OS, iOS, Android, Chrome OS, Windows OS (e.g., Windows XP, Windows Vista, Windows 7, Windows 8, Windows 10, Windows 11, Windows Server, etc.), Windows CE, Unix, Linux, SunOS, Solaris, Blackberry OS, VxWorks, or other suitable operating systems. In other embodiments, the computing device may be controlled by a proprietary operating system. Conventional operating systems control and schedule computer processes for execution, perform memory management, provide file systems, networking, I / O services, and provide user interface functions such as a graphical user interface ("GUI"), etc.

[0100] The references to "one example", "another example", "an example", etc. throughout this specification mean that the specific elements (e.g., features, structures, and / or characteristics) described in conjunction with the example are included in at least one example described herein, and may or may not be present in other examples. In addition, it should be understood that the elements used in any example may be combined in various examples in any suitable manner, unless the context clearly indicates otherwise.

[0101] It should be understood that the ranges provided herein include the specified range and any value or sub-range within the specified range, as if such value or sub-range was explicitly listed. For example, a range of about 2 kbp to about 20 kbp should be interpreted as including not only the explicitly listed limits of about 2 kbp to about 20 kbp, but also individual values, such as about 3.5 kbp, about 8 kbp, about 18.2 kbp, etc., and sub-ranges, such as about 5 kbp to about 10 kbp, etc. In addition, when "about" and / or "substantially" are used to describe a value, this means that a small variation (up to + / - 10%) of the stated value is included.

[0102] While several examples have been described in detail, it is to be understood that modifications may be made to the disclosed examples. Therefore, the above description should be considered non-limiting.

[0103] Although certain examples have been described, these examples are presented only as examples and are not intended to limit the scope of the present disclosure. In fact, the novel methods described herein can be embodied in various other forms. Moreover, various omissions, substitutions, and changes can be made in the methods described herein without departing from the spirit of the present disclosure. The attached claims and their equivalents are intended to cover such forms or modifications that will fall within the scope and spirit of the present disclosure.

[0104] Features, materials, characteristics or groups described in conjunction with a particular aspect or example are to be understood to be applicable to any other aspect or example described in this section or elsewhere in this specification unless incompatible therewith. All features disclosed in this specification (including any accompanying claims, abstract and drawings) and / or all steps of any method or process so disclosed may be combined in any combination, except combinations in which at least some of such features and / or steps are mutually exclusive. Protection is not limited to the details of any foregoing examples. Protection extends to any novel feature or any novel combination thereof of the features disclosed in this specification (including any accompanying claims, abstract and drawings), or to any novel step or any novel combination thereof of the steps of any method or process so disclosed.

[0105] Moreover, certain features described in the present disclosure in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented in multiple embodiments, either individually or in any suitable subcombination. Furthermore, although features may be described above as functioning in certain combinations, in some cases one or more features from the combination may be deleted from a claimed combination, and the combination may be claimed as a subcombination or a variation of a subcombination.

[0106] In addition, although operations may be depicted in the drawings or described in the specification in a particular order, it is not necessary to perform such operations or all operations in the particular order shown or in order to achieve the desired result. Other operations not depicted or described may be incorporated into the example methods and processes. For example, one or more additional operations may be performed before, after, simultaneously or between any described operations. Further, these operations may be rearranged or reordered in other embodiments. Those skilled in the art will appreciate that in some examples, the actual steps taken in the process shown and / or disclosed may be different from the actual steps shown in the figures. Depending on the example, some of the above steps may be removed, or other steps may be added. In addition, the features and attributes of the specific examples disclosed above may be combined in different ways to form additional examples, all of which fall within the scope of the present disclosure.

[0107] For purposes of the present disclosure, certain aspects, advantages, and novel features are described herein. Not all such advantages may be realized according to any particular example. Thus, for example, one skilled in the art will recognize that the present disclosure may be implemented or performed in a manner that achieves one advantage or group of advantages taught herein without necessarily achieving other advantages that may be taught or suggested herein.

[0108] Unless specifically stated otherwise or otherwise understood within the context of use, conditional language such as "can / could" or "might / may" is generally intended to convey that certain examples include and other examples do not include certain features, elements, and / or steps. Thus, such conditional language generally does not mean that one or more examples require the features, elements, and / or steps in any way, or that one or more examples must include logic for making decisions, with or without user input or prompting, whether or not features, elements, and / or steps are included in a particular example or are to be performed in any particular example.

[0109] Unless specifically stated otherwise, linking language such as the phrase "at least one of X, Y, and Z" is generally understood in context to convey that an item, term, etc. can be X, Y, or Z. Thus, such linking language is generally not intended to imply that certain examples require the presence of at least one of X, at least one of Y, and at least one of Z.

[0110] As used herein, language of degree, such as the terms "approximately," "about," "generally," and "substantially" represent a value, amount, or characteristic that comes close to the stated value, amount, or characteristic and still performs the intended function or achieves the intended result.

[0111] The scope of the present disclosure is not intended to be limited by the specific disclosure of preferred examples in this section or elsewhere in this specification, and may be limited by the claims presented in this section or elsewhere in this specification or presented in the future. The language of the claims should be interpreted broadly based on the language employed in the claims, and not limited to the examples described in this specification or during the prosecution of the application, which examples should be understood to be non-exclusive.

Claims

1. A computer-implemented method for training a machine learning model for detecting a biological component in a sample, the computer-implemented method comprising: Collect metagenomic data from biological components in the sample; generating a first molecular data set; generating a second molecular data set; creating a training set comprising an aggregated set of the first molecular dataset and the second molecular dataset; as well as The machine learning model is trained using the training set.

2. The method of claim 1, wherein the machine learning model comprises a random forest model.

3. The method of claim 1, wherein the machine learning model comprises a deep neural network (DNN).

4. The method of claim 1, wherein the machine learning model comprises a convolutional neural network (CNN).

5. The method of claim 1, wherein the machine learning model comprises a support vector machine (SVM).

6. The method of claim 1, wherein the machine learning model comprises a category classifier.

7. The method of any one of claims 1 to 6, further comprising selecting a first machine learning model based on one or more metrics of the first molecular dataset.

8. The method of claim 7, further comprising selecting a second machine learning model based on one or more metrics of the second molecular dataset.

9. The method of claim 8, wherein the first machine learning model and the second machine learning model are the same.

10. The method of claim 9, wherein the first machine learning model and the second machine learning model are different.

11. The method according to any one of claims 1 to 10, wherein the generating of the first molecular data set comprises applying an aligner-based classifier to the collected metagenomic data for a first source, and Wherein said generating of said second molecular dataset comprises applying said aligner-based classifier to said collected metagenomic data for a second source.

12. The method according to any one of claims 1 to 10, wherein the generating of the first molecular dataset comprises applying a de novo assembler to the collected metagenomic data for a first source, and Wherein said generating of said second molecular dataset comprises applying said de novo assembler to said collected metagenomic data for a second source.

13. The method of any one of claims 1 to 10, wherein the generating of the first molecular dataset comprises applying a k-mer based classifier to the collected metagenomic data for a first source, and Wherein the generating of the second molecular dataset comprises applying the k-mer based classifier to the collected metagenomic data for a second source.

14. The method according to any one of claims 1 to 10, wherein the generating of the first molecular dataset comprises applying a classifier to the collected metagenomic data for a first source, and Wherein said generating of said second molecular dataset comprises applying said classifier to said collected metagenomic data for a second source. The method of claim 14 , wherein the classifier comprises a category classifier. The method of claim 15 , wherein the classifier comprises a k-mer based classifier.

17. The method of any one of claims 11 to 16, wherein the first source comprises a first database.

18. The method of any one of claims 11 to 17, wherein the second source comprises a second database.

19. The method of any one of claims 11 to 16, wherein the first molecular dataset and the second molecular dataset comprise polypeptides.

20. The method of any one of claims 11 to 16, wherein the first molecular dataset and the second molecular dataset comprise polynucleotides.

21. The method of claim 20, wherein the first source comprises a curated collection of polynucleotides.

22. The method of claim 21, wherein the selected collection of polynucleotides comprises one or more genomes.

23. The method of any one of claims 20 to 22, wherein the polynucleotides of the second molecular dataset comprise publicly available polynucleotides.

24. The method of claim 23, wherein the publicly available polynucleotides comprise one or more publicly available genomes.

25. The method of claim 19, wherein the first source comprises a curated collection of polypeptides.

26. The method of claim 25, wherein the selected collection of polypeptides comprises one or more protein groups.

27. The method of any one of claims 19, 25 and 26, wherein the second source comprises a publicly available polypeptide.

28. The method of claim 27, wherein the publicly available polypeptides comprise one or more publicly available protein groups.

29. The method of any one of claims 1 to 28, wherein the first molecular dataset and the second molecular dataset comprise a plurality of taxids.

30. The method of claim 29, further comprising aggregating the first molecular dataset and the second molecular dataset for each of the taxids.

31. The method of claim 13 or 16, wherein the k-mer based classifier comprises a taxonomer.

32. The method of claim 13 or 16, wherein the k-mer based classifier comprises KRAKEN.

33. The method of any one of claims 1 to 32, further comprising detecting the presence of one or more of the biological components obtained from the sample based on a probability value from the output of the machine learning model using the training set.

34. The method of any one of claims 1 to 32, further comprising detecting the absence of one or more of the biological components obtained from the sample from the output of the machine learning model using the training set.

35. The method of any one of claims 1 to 34, wherein the sample is derived from one or more environmental sources, one or more industrial sources, one or more subjects, one or more microbial populations, or a combination thereof.

36. The method of claim 35, wherein the polynucleotides obtained from the sample include one or more polynucleotides from one or more pathogens.

37. The method of any one of claims 1 to 36, wherein the generating of the first molecular dataset and the second molecular dataset occurs in parallel.

38. The method of any one of claims 1 to 36, further comprising iterating the first molecular data set.

39. The method of any one of claims 1 to 38, further comprising iterating the second molecular data set.

40. A system for detecting a biological component in a sample, the system comprising: One or more processors programmed to perform a method comprising: obtaining metagenomic data, wherein the metagenomic data is obtained from biological components in the sample; generating a first molecular data set; generating a second molecular data set; creating a training set comprising an aggregated set of the first molecular dataset and the second molecular dataset; and The machine learning model is trained using the training set.

41. The system of claim 40, wherein the machine learning model comprises a random forest model.

42. The system of claim 40, wherein the machine learning model comprises a deep neural network (DNN).

43. The system of claim 40, wherein the machine learning model comprises a convolutional neural network (CNN).

44. The system of claim 40, wherein the machine learning model comprises a support vector machine (SVM).

45. The system of claim 40, wherein the machine learning model comprises a category classifier.

46. ​​The system of any one of claims 40 to 45, wherein the one or more processors are further programmed to perform a method comprising selecting a first machine learning model based on one or more metrics of the first molecular data set.

47. The system of claim 46, wherein the one or more processors are further programmed to perform a method comprising selecting a second machine learning model based on one or more metrics of the second molecular data set.

48. A system according to claim 47, wherein the first machine learning model and the second machine learning model are the same.

49. A system according to claim 47, wherein the first machine learning model and the second machine learning model are different.

50. The system of any one of claims 40 to 49, wherein the generating of the first molecular dataset comprises applying an aligner-based classifier to the collected metagenomic data for a first source, and Wherein said generating of said second molecular dataset comprises applying said aligner-based classifier to said collected metagenomic data for a second source.

51. The system of any one of claims 40 to 49, wherein the generating of the first molecular dataset comprises applying a de novo assembler to the collected metagenomic data for a first source, and Wherein said generating of said second molecular dataset comprises applying said de novo assembler to said collected metagenomic data for a second source.

52. The system of any one of claims 40 to 49, wherein the generating of the first molecular dataset comprises applying a k-mer based classifier to the collected metagenomic data for a first source, and Wherein the generating of the second molecular dataset comprises applying the k-mer based classifier to the collected metagenomic data for a second source.

53. The system of any one of claims 40 to 49, wherein the generating of the first molecular dataset comprises applying a classifier to the collected metagenomic data for a first source, and Wherein said generating of said second molecular dataset comprises applying said classifier to said collected metagenomic data for a second source.

54. The system of claim 53, wherein the classifier comprises a category classifier.

55. The system of claim 54, wherein the classifier comprises a k-mer based classifier.

56. A system according to any one of claims 50 to 55, wherein the first source comprises a first database.

57. A system according to any one of claims 50 to 56, wherein the second source comprises a second database.

58. The system of any one of claims 50 to 55, wherein the first molecular data set and the second molecular data set comprise polypeptides.

59. The system of any one of claims 50 to 55, wherein the first molecular data set and the second molecular data set comprise polynucleotides.

60. The system of claim 59, wherein the first source comprises a curated collection of polynucleotides.

61. The system of claim 60, wherein the selected collection of polynucleotides comprises one or more genomes.

62. The system of any one of claims 59 to 61, wherein the polynucleotides of the second molecular dataset comprise publicly available polynucleotides.

63. The system of claim 62, wherein the publicly available polynucleotides comprise one or more publicly available genomes.

64. The system of claim 58, wherein the first source comprises a curated collection of polypeptides.

65. The system of claim 64, wherein the selected collection of polypeptides comprises one or more protein groups.

66. The system of any one of claims 58, 64, and 65, wherein the second source comprises a publicly available polypeptide.

67. The system of claim 66, wherein the publicly available polypeptides comprise one or more publicly available protein groups.

68. The system of any one of claims 40 to 67, wherein the first molecular dataset and the second molecular dataset comprise a plurality of taxids.

69. The system of claim 68, wherein the one or more processors are further programmed to perform a method comprising aggregating the first molecular dataset and the second molecular dataset for each of the taxids.

70. The system of claim 52 or 55, wherein the k-mer based classifier comprises a taxonomer.

71. The system of claim 52 or 55, wherein the k-mer based classifier comprises KRAKEN.

72. A system according to any one of claims 40 to 71, wherein the one or more processors are also programmed to perform a method comprising detecting the presence of one or more of the biological components obtained from the sample based on a probability value from the output of the machine learning model using the training set.

73. A system according to any one of claims 40 to 71, wherein the one or more processors are further programmed to perform a method comprising detecting the absence of one or more of the biological components obtained from the sample from the output of the machine learning model using the training set.

74. The system of any one of claims 40 to 73, wherein the sample is derived from one or more environmental sources, one or more industrial sources, one or more subjects, one or more microbial populations, or a combination thereof.

75. The system of claim 74, wherein the polynucleotides obtained from the sample include one or more polynucleotides from one or more pathogens.

76. The system of any one of claims 40 to 75, wherein the generating of the first molecular dataset and the second molecular dataset occurs in parallel.

77. The system of any one of claims 40 to 75, wherein the one or more processors are further programmed to perform a method comprising iterating the generating of the first molecular data set.

78. The system of any one of claims 40 to 77, wherein the one or more processors are further programmed to perform a method comprising iterating the generating of the second molecular data set.

Images