Method for constructing species feature descriptors based on functional genomics
Through the species characteristic descriptor construction method based on functional genomics, the data integration problem of multi-species information in chemical toxicity prediction and risk assessment was solved, and the accurate description of inter-species differences and low-cost toxicity prediction model were achieved.
Patent Information
- Application Number
- CN202510562201.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-30
- Publication Date
- 2025-09-19
AI Technical Summary
Existing technologies make it difficult to effectively incorporate multi-species information into machine learning models for accurate chemical toxicity prediction and risk assessment, and existing species description methods face challenges such as difficulty in data collection and large differences between species.
A species characteristic descriptor construction method based on functional genomics is adopted. By collecting proteomic data of target species, using software tools for functional gene annotation and visualization, multi-species descriptors are constructed, including characteristic descriptors of molecular functions, biological processes and protein categories, and cloud servers are used for data management and visualization.
It achieves accurate capture of interspecies differences, reduces information processing costs, and improves the accuracy and efficiency of multi-species chemical toxicity prediction and risk assessment.
Smart Images
Figure CN120673838A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of toxicity prediction and risk assessment modeling of chemicals to multiple species, and in particular to a method for constructing species characteristic descriptors based on functional genomics. Background Art
[0002] The rapid emergence of new chemicals poses a major challenge to regulatory agencies in conducting comprehensive risk assessments. Due to the complexity of the toxic mechanisms of chemicals, elucidating their toxic mechanisms through in vivo and in vitro experimental methods can be very tedious and challenging. Therefore, there is an increasing need to combine computer simulation methods to supplement experimental work in order to more accurately predict the toxicological effects of new chemicals and enhance our understanding of their toxicological mechanisms. The emergence of machine learning (ML) has provided an effective method for ecotoxicological risk assessment using computer methods. However, the actual application scenarios of the model require accurate toxicity predictions for multiple species, and we still lack methods that incorporate multi-species information into the model to achieve predictive multi-species chemical risk assessment methods. Therefore, we need to develop a set of descriptors that can accurately capture differences between species.
[0003] Recent studies have successfully used machine learning (ML) models to explore the relationships between the physicochemical properties of chemicals, their environmental exposure conditions, and their related toxicological endpoints. However, research on how to incorporate multi-species information into these models is still insufficient. Current studies at home and abroad have attempted to use species trophic level and species physical size information as methods to describe interspecies differences, but these methods face challenges such as data collection difficulties and large interspecies differences. Although current research has made initial progress in analyzing multi-species, it shows that incorporating multi-species information into modeling is beneficial to toxicological prediction models. However, how to effectively and appropriately integrate multi-species data into these models remains a challenge. Therefore, it is necessary to develop a set of descriptors that can accurately capture interspecies differences and meet the following criteria: (1) can fully reflect the differences between species, (2) are easy to be machine-readable (minimize information processing costs), and (3) are relatively easy to obtain and have high data integrity. Summary of the Invention
[0004] The present invention addresses the deficiencies in the prior art and provides a method for constructing species characteristic descriptors based on functional genomics, which is beneficial for providing a new method for toxicity prediction and risk assessment modeling of chemicals to multiple species.
[0005] The technical solutions of the present invention are as follows:
[0006] A method for constructing species characteristic descriptors based on functional genomics, characterized by comprising the following steps:
[0007] Step 1: Collect proteomic data of the target species;
[0008] Step 2: deploying a software tool for extracting species functional gene information on a cloud server, and developing a command line tool for the software tool to enable gene annotation of multiple species and upload and download of data;
[0009] Step 3, importing the proteomic data into the tool software for query, matching and annotation to obtain the functional genomic information of the target species;
[0010] Step 4: preprocess and visualize the annotated functional genomic information;
[0011] Step 5: Construct species characteristic descriptors for multiple species based on functional genomic information.
[0012] Step 5 includes incorporating the ratio of the number of functional genes of a specific function in the category to which the function of the target species belongs to the total number of genes of the target species into the descriptor of the specific function based on the annotated functional genomic information. For specific functions that the target species lacks but exists in the multi-species gene function summary table, the ratio in its descriptor is marked as zero. For specific functions of the target species that exceed the multi-species gene function summary table and exist only in a negligible number, they are excluded from the descriptor, thereby utilizing all descriptors for gene function modeling.
[0013] The functional categories of the multi-species gene function summary table include molecular function category, biological process category and protein category.
[0014] The molecular function category includes the following specific functions: transporter activity, ATP-dependent activity, translation regulator activity, molecular converter activity, cargo receptor activity, molecular adaptor activity, transcriptional regulatory activity, antioxidant capacity, catalytic activity, structural molecule activity, molecular function regulator, binding capacity, and cytoskeletal motor activity.
[0015] The biological process category includes the following specific functions: cellular processes, multicellular biological processes, reproductive processes, locomotion and positioning functions, metabolic processes, biological processes involved in interspecies interactions, growth, detoxification, immune system processes, reproduction, rhythmic processes, biological regulation, biological stages, stress response, biological adhesion, homeostatic processes, signal transduction, pigmentation, biomineralization, and developmental processes.
[0016] The protein classes include the following specific functions: extracellular matrix proteins, translation proteins, cytoskeletal proteins, metabolite interconversion enzymes, transporters, protein modifying enzymes, scaffold / adaptation proteins, chromatin / chromatin binding or regulatory proteins, DNA metabolic proteins, transport / carrier proteins, cell adhesion molecules, membrane trafficking proteins, intercellular signaling molecules, chaperone proteins, protein binding activity regulators, cell link proteins, viral or locus element proteins, structural proteins, RNA metabolic proteins, storage proteins, calcium binding proteins, transmembrane signaling receptors, gene-specific transcriptional regulators, and defense / immunity proteins.
[0017] The proteomic data in step 1 includes protein sequence files.
[0018] The software tool in step 2 is capable of functional annotation by matching input protein sequences with pre-annotated phylogenetic trees, and has an integrated workflow for managing multi-species genomic data and effectively visualizing results, thus enabling seamless data management, multi-species genome annotation, and result visualization.
[0019] The functional genomic information in step 3 includes gene number information.
[0020] The technical effects of the present invention are as follows: The present invention provides a method for constructing species characteristic descriptors based on functional genomics, which utilizes species characteristic descriptors formed based on functional genomics, a basic species information, to facilitate both accurate description of differences between species and quantitative assessment related to the toxic effects of chemical substances in organisms.
[0021] The present invention can achieve the following features: constructing the functional genomic information of the target species based on the proteomic data of the species, completing the annotation of the proteomic data by developing and deploying software tools; classifying and visualizing the functional genomic information of the species, extracting the proportion of functional genes to the total number of genes, and constructing a multi-species descriptor.
[0022] Digitize exposure environment and species diversity information so that it can be modeled; use different algorithms and compare the results, and perform internal and external validation of the model; analyze the importance of features and the interactions between features to identify the most influential features and the relationships between important features.
[0023] The present invention can accurately capture descriptors of species differences while meeting the following criteria: (1) fully reflecting the differences between species; (2) easily machine-readable (minimizing information processing costs); and (3) relatively easy to obtain and high data integrity. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1It is a flow chart of a method for constructing species characteristic descriptors based on functional genomics according to the present invention. Figure 1 The process includes step 1, collecting proteomic data of species; step 2, developing and deploying software tools; step 3, querying, matching and annotating the data; step 4, extracting and visualizing functional genomic information; and step 5, preprocessing and constructing species descriptors. DETAILED DESCRIPTION
[0025] Below is the attached figure ( Figure 1 ) and Examples illustrate the present invention.
[0026] Figure 1 This is a flow chart of a method for constructing species characteristic descriptors based on functional genomics. Figure 1 As shown, a method for constructing species characteristic descriptors based on functional genomics includes the following steps: Step 1, collecting proteomic data of the target species; Step 2, deploying a software tool for extracting species functional gene information on a cloud server, and developing a command line tool for the software tool to realize gene annotation of multiple species and upload and download of data; Step 3, importing the proteomic data into the tool software for query, matching and annotation to obtain the functional genomic information of the target species; Step 4, preprocessing and visualizing the annotated functional genomic information; Step 5, constructing species characteristic descriptors for multiple species based on the functional genomic information.
[0027] Step 5 includes incorporating the ratio of the number of functional genes of a specific function in the category to which the function of the target species belongs to the total number of genes of the target species into the descriptor of the specific function based on the annotated functional genomic information. For specific functions that the target species lacks but exists in the multi-species gene function summary table, the ratio in its descriptor is marked as zero. For specific functions of the target species that exceed the multi-species gene function summary table and exist only in a negligible number, they are excluded from the descriptor, thereby utilizing all descriptors for gene function modeling.
[0028] The functional categories of the multi-species gene function summary table include molecular function categories, biological process categories, and protein categories. The molecular function categories include the following specific functions: transporter activity, ATP-dependent activity, translation regulator activity, molecular converter activity, cargo receptor activity, molecular adaptor activity, transcriptional regulation activity, antioxidant capacity, catalytic activity, structural molecule activity, molecular function regulator, binding capacity, and cytoskeletal motility activity.
[0029] The biological process category includes the following specific functions: cellular processes, multicellular biological processes, reproductive processes, locomotion and positioning functions, metabolic processes, biological processes involved in interspecies interactions, growth, detoxification, immune system processes, reproduction, rhythmic processes, biological regulation, biological stages, stress response, biological adhesion, homeostatic processes, signal transduction, pigmentation, biomineralization, and developmental processes. The protein category includes the following specific functions: extracellular matrix proteins, translation proteins, cytoskeletal proteins, metabolite interconversion enzymes, transporters, protein modifying enzymes, scaffold / adaptive proteins, chromatin / chromatin binding or regulatory proteins, DNA metabolic proteins, transport / carrier proteins, cell adhesion molecules, membrane trafficking proteins, intercellular signaling molecules, chaperone proteins, protein binding activity regulators, cell link proteins, viral or locus element proteins, structural proteins, RNA metabolic proteins, storage proteins, calcium binding proteins, transmembrane signaling receptors, gene-specific transcriptional regulators, and defense / immune proteins.
[0030] The proteomic data in step 1 includes protein sequence files. The software tool in step 2 performs functional annotation by matching input protein sequences to pre-annotated phylogenetic trees. It features an integrated workflow for managing multi-species genomic data and effectively visualizing results, enabling seamless data management, multi-species genome annotation, and visualization. The functional genomic information in step 3 includes gene ID information.
[0031] refer to Figure 1 The present invention works as follows: The first step is to collect species proteomic data, followed by the deployment of annotation software and the development of command-line tools. After obtaining the species' functional gene information, it is classified and visualized, screened, and preprocessed. Finally, a multi-species descriptor based on functional genomics is constructed, providing a new approach for predicting the toxicity of chemicals on multiple species and for modeling risk assessment. Compared to other species description methods for chemical toxicity prediction and risk assessment models, this approach is distinguished by the development of a set of quantitative descriptors based on the functional genomics—essential species information—that accurately describe differences between species and correlate with the toxic effects of chemicals in organisms.
[0032] The present invention will construct a set of multi-species descriptors that can accurately describe the differences between species based on species functional genomic information. To this end, we have adopted the following technical solution, which is divided into five steps: (1) Collecting proteomic data of the target species; (2) Developing software tools to extract species functional gene information; (3) Importing proteomic data into the software for query, matching and annotation to obtain the functional genomic information of the species; (4) Preprocessing and visualizing the annotated functional genomic information; (5) Constructing multi-species descriptors based on functional genomic information. The following are the specific method steps of this patent:
[0033] I. Collection of genomic data
[0034] The development of species descriptors begins with collecting functional gene information for each species. However, for aquatic organisms, only a few species currently have complete functional gene data. Therefore, it is necessary to independently obtain functional gene information for aquatic organisms. We searched and downloaded protein sequence files (FAA format) for 31 target species from the National Center for Biotechnology Information (NCBI) database. Species can be categorized into the following groups: algae (5 species), bacteria (10 species), fish (5 species), daphnia (3 species), plants (2 species), benthos (3 species), and protists (3 species).
[0035] 2. Develop software tools to extract functional gene information from species
[0036] To annotate gene functions in a single species, Tang et al. proposed the software tool TreeGrafter, which performs functional annotation by matching input protein sequences with pre-annotated phylogenetic trees. However, it does not provide an integrated workflow for managing multi-species genomic data or effectively visualizing the results. To address this gap, we developed a comprehensive workflow that extends the functionality of TreeGrafter by enabling seamless data management, multi-species genome annotation, and visualization of the results: we deployed TreeGrafter on a cloud server and developed a command-line tool to process these sequences using TreeGrafter, enabling annotation of genes from multiple species and uploading and downloading of data.
[0037] 3. Acquisition of species functional genomic information
[0038] Upload the species proteome data obtained in step 1 to the cloud server and input the TreeGrafter software deployed in step 2. After the software completes the functional annotation, it outputs the name and function number information of each gene. Finally, convert the data from the text file into xlsx format and download it locally.
[0039] IV: Preprocessing and visualization of functional genomic information
[0040] The annotated functional genomic information was processed using the bioinformatics analysis website PANTHER, which provides functional gene information based on various classification methods. In addition to Gene Ontology (GO)-based classifications (such as molecular function (MF), biological process (BP), and cellular component (CC)), it also includes protein category (PC) and pathway classification information. Protein category is a method of classifying different proteins based on characteristics such as function, cellular location, evidence, or disease association. We selected MF, BP, and PC as the information sources for constructing species descriptors because they are closely related to biological processes and cover gene functions highly relevant to toxic effects, such as oxidative stress, metabolism, and detoxification. Entering the gene number information for a species from step 3 into the online tool on the PANTHER website will generate an xlsx file containing the number of functional genes for each of the three classification methods, MF, BP, and PC, and the proportion of these functional genes in the total number of functional genes.
[0041] We collected a total of 55 different gene functions from 31 species. According to the classification method of Gene Ontology, gene functions can be divided into three categories: 12 types of molecular functions (MF), 19 types of biological processes (BP) and 24 types of protein classes (PC) (Table 1).
[0042] Table 1: Overview of gene functions in species
[0043]
[0044]
[0045] 5. Construction of multi-species descriptors based on functional genomic information
[0046] After obtaining the functional gene information for each target species, it is necessary to convert it into machine-readable descriptors. Because most species share similar basic gene functions, interspecies differences are primarily reflected in variations in the number of genes with specific functions. Therefore, based on the functional gene information obtained in step 4, we use the ratio of each specific functional gene type relative to the total gene count for each species as a descriptor. When a particular species lacks a gene function present in other species, the gene function is marked as zero. In addition, when certain gene functions are absent in all species or present only in negligible quantities, they are excluded from the descriptors, and the remaining gene functions are selected for modeling. The final number of functional genes used for modeling is 50, covering 12 molecular functions, 17 biological processes, and 21 protein categories.
[0047] Taking zebrafish as an example, Table 2 shows the proportion of each functional gene in the total number of functional genes of this species:
[0048] Table 2: Zebrafish gene descriptors
[0049]
[0050]
[0051] Any content not described in detail in this specification is prior art known to those skilled in the art. It should be noted that the above description is intended to help those skilled in the art understand the present invention, but does not limit the scope of protection of the present invention. Any equivalent substitution, modification, improvement, and / or simplification of the above description that does not depart from the essence of the present invention shall fall within the scope of protection of the present invention.
Claims
1. A method for constructing species characteristic descriptors based on functional genomics, characterized in that: The following steps are involved: Step 1: Collect proteomic data of the target species; Step 2: deploying a software tool for extracting species functional gene information on a cloud server, and developing a command line tool for the software tool to enable gene annotation of multiple species and upload and download of data; Step 3, importing the proteomic data into the tool software for query, matching and annotation to obtain the functional genomic information of the target species; Step 4: preprocess and visualize the annotated functional genomic information; Step 5: Construct species characteristic descriptors for multiple species based on functional genomic information.
2. The method for constructing species characteristic descriptors based on functional genomics according to claim 1, characterized in that: Step 5 includes incorporating the ratio of the number of functional genes of a specific function in the category to which the function of the target species belongs to the total number of genes of the target species into the descriptor of the specific function based on the annotated functional genomic information. For specific functions that the target species lacks but exists in the multi-species gene function summary table, the ratio in its descriptor is marked as zero. For specific functions of the target species that exceed the multi-species gene function summary table and exist only in a negligible number, they are excluded from the descriptor, thereby utilizing all descriptors for gene function modeling.
3. The method for constructing species characteristic descriptors based on functional genomics according to claim 2, characterized in that: The functional categories of the multi-species gene function summary table include molecular function category, biological process category and protein category.
4. The method for constructing species characteristic descriptors based on functional genomics according to claim 3, characterized in that: The molecular function category includes the following specific functions: transporter activity, ATP-dependent activity, translation regulator activity, molecular converter activity, cargo receptor activity, molecular adaptor activity, transcriptional regulatory activity, antioxidant capacity, catalytic activity, structural molecule activity, molecular function regulator, binding capacity, and cytoskeletal motor activity.
5. The method for constructing species characteristic descriptors based on functional genomics according to claim 3, characterized in that: The biological process category includes the following specific functions: cellular processes, multicellular biological processes, reproductive processes, locomotion and positioning functions, metabolic processes, biological processes involved in interspecies interactions, growth, detoxification, immune system processes, reproduction, rhythmic processes, biological regulation, biological stages, stress response, biological adhesion, homeostatic processes, signal transduction, pigmentation, biomineralization, and developmental processes.
6. The method for constructing species characteristic descriptors based on functional genomics according to claim 3, characterized in that: The protein classes include the following specific functions: extracellular matrix proteins, translation proteins, cytoskeletal proteins, metabolite interconversion enzymes, transporters, protein modifying enzymes, scaffold / adaptation proteins, chromatin / chromatin binding or regulatory proteins, DNA metabolic proteins, transport / carrier proteins, cell adhesion molecules, membrane trafficking proteins, intercellular signaling molecules, chaperone proteins, protein binding activity regulators, cell link proteins, viral or locus element proteins, structural proteins, RNA metabolic proteins, storage proteins, calcium binding proteins, transmembrane signaling receptors, gene-specific transcriptional regulators, and defense / immunity proteins.
7. The method for constructing species characteristic descriptors based on functional genomics according to claim 1, characterized in that: The proteomic data in step 1 includes protein sequence files.
8. The method for constructing species characteristic descriptors based on functional genomics according to claim 1, characterized in that: The software tool in step 2 is capable of functional annotation by matching input protein sequences with pre-annotated phylogenetic trees, and has an integrated workflow for managing multi-species genomic data and effectively visualizing results, thus enabling seamless data management, multi-species genome annotation, and result visualization.
9. The method for constructing species characteristic descriptors based on functional genomics according to claim 1, characterized in that: The functional genomic information in step 3 includes gene number information.