Computer-implemented method for generating comparative proteomic analysis reports from multiple biological samples

Abstract

The invention relates to a computer-implemented method for generating a proteomic analysis report for comparing a first biological sample (X1) having undergone a treatment and a control sample, characterised in that it includes the following steps: (E1) receiving data relating to proteins present in each biological sample, comprising values relating to the expression level of the proteins in each sample; (E2) grouping the data into groups containing the data relating to the proteins present in each sample and involved in the same biological process; (E3) calculating, from the values relating to the expression levels, comparative values of the expression levels of the first sample to the control sample; and (E4) generating a proteomic analysis report including a first graphic indicating a biological process and a graphic element indicating the comparative value of this biological process.

Description

Description Title of the invention: Computer-implemented method for generating comparative proteomic analysis reports from multiple biological samples Technical Field

[0001] The present invention relates to the technical field of proteomic analysis. More specifically, it relates to automated methods for processing, comparing and visualizing proteomic data from multiple biological samples. Introduction

[0002] In the above field, it is known to use advanced techniques such as mass spectrometry to obtain detailed proteomic data. These techniques have made it possible to identify and quantify thousands of proteins present in biological samples, thus providing an overview of cellular processes and metabolic pathways. However, despite the wealth of data obtained, their interpretation and analysis remain a major challenge. Traditional methods have often relied on manual or semi-automated analyses, requiring significant expertise and being prone to human error.

[0003] Additionally, comparing proteomic profiles between different samples, such as treated and control samples, was often performed without effective visualization tools. Researchers found themselves overwhelmed by the amount of data, making it difficult to detect meaningful trends or patterns. Previous methods also did not provide effective ways to group proteins based on their biological functions or metabolic pathways, which could have facilitated interpretation.

[0004] Furthermore, while the state of the art recognized the importance of comparing protein expression levels to a control sample, it did not provide systematic methods to do so in an automated manner. Researchers often had to rely on third-party software or in-house algorithms to make these comparisons, which could introduce inconsistencies and errors.

[0005] In short, there is a gap in the field of proteomic analysis regarding the automated analysis, comparison, and visualization of this data in a consistent and efficient manner. This lack has hampered researchers' ability to draw accurate conclusions and move forward quickly in their studies. Detailed description

[0006] There is therefore a need for an automated and integrated solution capable of efficiently processing, comparing and visualizing proteomic data from multiple samples, identifying specific biological processes by grouping functionally related proteins, and presenting this information in a synthetic and understandable visual form, while eliminating the drawbacks and limitations associated with traditional state-of-the-art methods.

[0007] The invention is placed in this context and seeks to resolve the aforementioned drawbacks, by proposing a method for generating a computer-implemented proteomic analysis report for the comparison of two biological samples, the first biological sample having undergone treatment and the second biological sample being a control sample, the method comprising the following steps: a. a step of receiving data relating to a plurality of proteins present in each biological sample, comprising: for each protein of said plurality of proteins, an identifier of said protein and at least one value relating to the expression level of this protein in each sample; b.a step of grouping the data into data groups using at least one database containing a list of proteins, each associated with an identifier and a metabolic pathway linked to a biological process; each data group containing the data relating to the proteins present in each biological sample and involved in the same biological process; c. for each data group, a step of calculating, from the values ​​relating to the expression rate, at least one comparative value of the expression rates of the proteins of this group in the first sample compared to the expression rates of the proteins of this group in the control sample; d.a step of generating a proteomic analysis report comprising at least a first graph indicating at least one biological process associated with a group of data and, for each biological process indicated on this graph, at least one graphic element indicating said comparative value obtained for the group of data associated with this biological process.

[0008] Advantageously, the method of generating a computer-implemented proteomic analysis report for the comparison of two biological samples allows an analysis accurate and automated analysis of proteomic differences between at least one treated sample and a control sample, the control sample being able to be a sample of the same nature as the treated sample but having not undergone any treatment, the control sample then being called an untreated sample, or having undergone a neutral treatment, the control sample then being called a vehicle. The first step of receiving data relating to a plurality of proteins present in each biological sample provides a comprehensive view of the expressed proteins, thus facilitating a detailed comparison. This is particularly useful for identifying subtle variations in protein expression that could be the consequence of a treatment.

[0009] For each protein present and for each sample, a said value relating to the expression level of this protein in each sample is indicated. This value is for example determined from a count of the peptides composing this protein in the sample. In the case where a protein is detected without its expression level being able to be determined, said value may be estimated from a detection limit value of the system used to generate said received data. For example, in the case where the method is implemented for triplicates and data relating to one of the proteins is missing for one of the samples in the triplicate, this data may be calculated from the existing data of the other samples.

[0010] This value may be a relative value, such as a ratio between the expression rate of this protein in the first sample, or the average of the expression rates of this protein in the first samples of the same group of samples having undergone the same treatment, and the expression rate of this protein in the second sample, or the average of the expression rates of this protein in the second samples of the same group of control samples. Alternatively or cumulatively, this value may be an absolute value, such as the expression rate of this protein in the first sample, or the average of the expression rates of this protein in the first samples of the same group of samples having undergone the same treatment.It is conceivable that the data include several values, and in particular for each protein: ratios of expression rates between each sample of the same group of samples having undergone the same treatment and between each sample of the same group of control samples; averages of these expression rates on these groups of samples as well as one or more statistical values, for example representative of the statistical significance of these values, such as a p-value.

[0011] The step of grouping the data into data groups using a database containing a list of proteins associated with specific metabolic pathways allows to contextualize protein expression changes within broader biological processes. This offers a major advantage in allowing users to understand not only which proteins are affected by a treatment, but also how these changes can influence entire metabolic pathways.

[0012] The step of calculating a comparative value of expression levels for each data group provides a quantitative measure of the differences between samples. This allows for an objective assessment of the effects of a treatment on protein expression, thus facilitating informed decisions about the relevance or effectiveness of a treatment.

[0013] Finally, the step of generating a proteomic analysis report with a graph indicating the biological processes and the associated comparative values ​​makes the results easily accessible and understandable. It is then easy to quickly identify the biological processes most affected by a treatment, which is essential for decision-making and planning future studies.

[0014] During this step, mass spectrometry is used to detect the peptides present in each biological sample. These detected peptides then serve as markers to identify the corresponding proteins. Each protein is identified by associating the detected peptides with a database containing a list of proteins, each associated with a unique identifier. Once the proteins are identified, the abundance ratios are calculated. These ratios represent the quantitative relationship between the proteins present in the two biological samples compared. If several samples with the same treatment and several control samples are present, the average of the 1000 log2 of the ratios of the average expression levels of the former to the average expression levels of the latter is calculated. This comparative value of the expression levels makes it possible to obtain a standardized measure of the variation in protein expression between the samples.

[0015] In one embodiment of the invention, the method culminates with a step dedicated to saving and disseminating the generated proteomic analysis report. In this final phase, several actions can be undertaken simultaneously or sequentially.

[0016] First, the report is saved digitally. This backup ensures that all valuable data and analysis are preserved for future reference, review, or further analysis.

[0017] Second, the report can be transmitted to a remote digital terminal. Remote transmission ensures rapid and efficient dissemination of information, enabling real-time discussions, evaluations, or decision-making.

[0018] Finally, it's also possible to print the report on paper. Although we live in the digital age, having a paper version available can be invaluable for presentations, meetings, or simply for those who prefer a tangible reading experience. Printing offers additional flexibility in how information can be accessed and shared.

[0019] In the context of the present invention, the term "abundance ratio" means the quantitative ratio between the proteins present in two distinct biological samples. This measurement is often expressed in logarithmic terms, in particular using Log2, to facilitate the comparison and interpretation of variations in protein expression between the samples. For a given protein, said ratio may be estimated by calculating ratios between the abundances of said protein in said samples, or differences between said abundances in the case where they are expressed in logarithmic terms. These ratios make it possible in particular to calculate a ratio of the average level of expression of a protein between a sample subjected to a condition and another sample, such as a control sample. This ratio, also called "fold change", thus makes it possible to translate the change in the expression of the protein under the influence of the condition.

[0020] In the context of the present invention, "mass spectrometry" means an analytical technique for identifying and quantifying molecules based on their mass and charge. It is commonly used to detect and identify peptides and proteins in biological samples, thus providing essential information for proteomic analysis.

[0021] In the context of the present invention, "unique identifier" means a specific code or sequence assigned to each protein listed in a database. This identifier makes it possible to distinguish and trace each protein uniquely, thus facilitating its analysis and comparison in different samples.

[0022] In the context of the present invention, "metabolic pathway" means a series of chemical interactions occurring within a cell, generally involving several proteins and other molecules, to convert one substance into another or one signal into another. These pathways are essential for understanding biological processes and potential dysfunctions associated with certain diseases or conditions.

[0023] In a cumulative embodiment of the invention, the method for generating a proteomic analysis report comprises a prior step of implementing a mass spectrometry on at least two biological samples to generate said data relating to a plurality of proteins present in each biological sample.

[0024] Advantageously, the integration of a preliminary step of implementing mass spectrometry on both biological samples considerably enriches the process of generating a proteomic analysis report. Mass spectrometry is a powerful and precise technique for identifying and quantifying proteins in a biological sample. By using this method to generate data relating to a plurality of proteins present in each sample, the method guarantees reliable detection and quantification of proteins and / or peptides, even at very low concentrations.

[0025] By doing so, it is possible to obtain high-quality proteomic data, which is essential for accurate comparison between samples. In addition, mass spectrometry can detect post-translational modifications, including those induced by enzymes, and protein isoforms induced by alternative splicing, thus providing a more comprehensive and detailed view of the proteomics of the samples. As a result, the results obtained are not only quantitative but also qualitative, allowing a better understanding of the proteomic changes between treated and control samples.

[0026] Alternatively, instead of exclusively using mass spectrometry to generate data relating to a plurality of proteins present in each biological sample, other proteomic techniques may be employed. For example, liquid chromatography coupled with mass spectrometry may be used so that the peptides resulting from the enzymatic hydrolysis of proteins are initially separated by reverse-phase liquid chromatography according to their hydrophobicity and thus separate and identify complex proteins, ensuring better quantification of the proteins.

[0027] If desired, in addition to mass spectrometry, fluorescence microscopy can be used to visualize the location and abundance of specific proteins in biological samples, thus providing contextual information on the cellular localization of proteins.

[0028] Alternatively, microarray-based proteomics can be implemented, allowing the detection of molecular interactions with other proteins, peptides, nucleic acids, and other structures. This approach is particularly useful for high-throughput studies where many samples need to be analyzed quickly.

[0029] Advantageously, techniques such as electrospray ionization mass spectrometry or MALDI-TOF mass spectrometry can be used, particularly to analyze chemical modifications of proteins.

[0030] In one embodiment of the invention, each protein listed in the database is associated with a unique identifier; in addition, the prior step of implementing mass spectrometry comprises the detection of peptides present in each biological sample, the identification of proteins present in each biological sample from the identified peptides and the association with each identified protein of one of said unique identifiers.

[0031] Advantageously, associating each protein listed in the database with a unique identifier greatly simplifies the process of identifying and comparing proteins between different biological samples. This unique identifier ensures that each protein is clearly recognized, thus avoiding ambiguities or confusion that could arise during data analysis. It also facilitates the integration and comparison of data between different studies or experiments, as each protein is clearly and uniformly identified.

[0032] The unique identifier associated with each protein may be an identifier specific to the mass spectrometry used in the previous step, such as a Uniprot code. This identifier may be distinct from a unique identifier associated with the same protein in another database, such as a STRING code.

[0033] The preliminary step of implementing mass spectrometry, which involves the detection of peptides present in each biological sample, offers increased accuracy in protein identification. By focusing first on peptides, the method allows for more sensitive detection of proteins, including those present in low quantities. In addition, identifying proteins from the detected peptides ensures that proteins are identified based on their elementary constituents, thus increasing the reliability of identification.

[0034] Associating each identified protein with one of these unique identifiers further enhances the accuracy and efficiency of the process. Once a protein is identified from its peptides, it is immediately associated with its unique identifier. The unique identifier serves as a universal reference for each protein, thus ensuring clear traceability and efficient data organization.

[0035] In a cumulative embodiment of the invention, the data relating to a plurality of proteins present in each biological sample comprises, for each protein identified and regardless of the method of identification and assignment of peptides, the number of detected peptides associated with it.

[0036] It may be provided that, for each peptide and / or each protein identified in each sample, said data include a statistical value associated with this peptide and / or this protein, such as a p-value or a false discovery rate or FDR.

[0037] In one embodiment of the invention, the method comprises a filtering step, prior to the grouping step, in which the data relating to the proteins present in each biological sample and comprising a single unique peptide are removed from said data.

[0038] Advantageously, including data on the number of peptides detected and composing each identified protein helps ensure a high degree of accuracy in the analysis, regardless of the method chosen. By knowing the number of peptides that compose a specific protein, it is possible to have a better idea of ​​the coverage and confidence of the identification of that protein. This can help distinguish between similar proteins and confirm the identification of less abundant proteins or proteins that might be difficult to detect.

[0039] The filtering step, prior to the clustering step, where data associated with each protein with a unique peptide is removed, is particularly advantageous for improving the reliability of the analysis. Proteins identified by a single peptide can often be considered less reliable or potentially erroneous identifications. By eliminating this data, the process ensures that only proteins identified with a higher degree of confidence are considered in the final analysis. This reduces the risk of errors or false identifications, ensuring that the proteomic analysis report is based on robust and reliable data.

[0040] If desired, instead of simply removing data for proteins containing a single peptide, a validation step could be included. This validation step could use other techniques or methods to confirm the identification of these proteins, such as additional analyses or bioinformatics validation techniques. This would notably ensure that certain rare or less abundant, but nevertheless relevant, proteins are not prematurely discarded.

[0041] Alternatively or cumulatively, the method comprises another filtering step, prior to the grouping step, in which only the data relating to the proteins present in the first biological sample and whose expression rate is increased by significantly compared to the second sample are removed from said data. Said filter could for example be implemented by comparing the abundance ratio of each protein between the first sample and the second sample to a given threshold value, predetermined or set in advance by a user, and by comparing the p-value associated with this protein to another given threshold value. It will thus be possible to retain only the data relating to proteins whose expression is increased and / or proteins whose expression is decreased.

[0042] Additionally, in addition to filtering based on the number of peptides, other criteria could be used to improve data quality. For example, a threshold based on signal intensity or spectral quality could be applied to eliminate protein identifications with weak or poor quality signals.

[0043] As an alternative to simply counting the number of peptides composing each protein, a more in-depth analysis of peptide motifs or specific sequences can be performed. This analysis can provide additional information on post-translational modifications or protein isoforms, leading to a more detailed analysis of the protein composition of each sample.

[0044] Advantageously, a machine learning system can be integrated to identify and eliminate false positives or to recognize recurring patterns in the data, thus improving the accuracy and reliability of proteomic analysis.

[0045] In one embodiment of the invention, and preferably subsequent to the filtering step and prior to the grouping step, the method comprises a step of generating missing data in which, for at least one of said plurality of proteins present in one of the samples for which said at least one value relating to the expression level of this protein in this sample is missing in said data, an estimate of said value is calculated from the value relating to the expression level of this protein in at least one other sample.Where appropriate, in the case where the method is implemented for a group of first samples having undergone the same treatment, in particular a triplicate, and one of said values ​​relating to the expression rate of one of the proteins is missing for one of the samples of said group, in particular for only one of the samples of said group, an estimate of said value may be calculated from the existing values, relating to the expression rate of this protein in the other samples of said group.

[0046] It may indeed happen that data relating to a protein are missing for one or more samples in the same group of samples, given the random sampling of a spectrometer and its detection limit. It is then common not to take into account all the data relating to this protein. This strategy can thus lead to losing a significant percentage of the data acquired during spectrometry. All of the characteristics mentioned above thus make it possible to achieve a correct estimation of the missing values, in particular by replacing them with values ​​representing the detection limit and thus to robustify the proteomic analysis.

[0047] Advantageously, said step of generating missing data may comprise, for each protein of which one of said values ​​is missing in said data: a sub-step of normalization of the existing values ​​in said data and relating to the expression level of this protein in the samples of said group; and a sub-step of random selection of a value according to a statistical distribution, in particular normal, corresponding to an interval of less than 2.5% of the distribution of the normalized values. The existing values ​​may in particular be normalized using their mean and their correlation to the mean. Reference may in particular be made to the publication “Guide for protein fold change and p-value calculation for non-experts in proteomics” (Aguilan, K. Kulej and S. Sidoli, Mol. Omics, 2020, DOI: 10.1039 / D0MO00087F) which describes such a step of generating missing data.

[0048] Said estimates of missing values ​​can thus make it possible to calculate ratios of expression rates between each sample of the same group of samples having undergone the same treatment and between each sample of the same group of control samples; averages of these expression rates over these groups of samples, ratios of these averages or any other value of the “fold change” type likely to reflect the change in the expression of the protein under the influence of a condition compared to a control sample.

[0049] Where appropriate, said data generation step may include a sub-step of calculating a statistical value associated with this missing value estimate, such as a p-value or a false discovery rate or FDR.

[0050] In an alternative or cumulative embodiment of the invention, the step of grouping the data into data groups comprises a step of identifying, using at least one database containing a list of proteins each associated with an identifier and at least one interaction with another protein in this list, said interaction being involved in at least one metabolic pathway linked to a biological process, interactions of the proteins present in each biological sample with each other, and a step of generating said data group from said identified interactions.

[0051] Advantageously, the integration of the step of identifying the interactions of the proteins present in each biological sample with each other considerably enriches the depth of the proteomic analysis. By identifying these interactions, the method is not limited to a simple list of proteins present, but also offers an overview of the functional networks and metabolic pathways active in the sample. This allows a comprehensive understanding of the biological state of the sample by highlighting the joint functioning of the expressed proteins.

[0052] Using at least one database containing a list of proteins, each associated with an identifier and at least one interaction with another protein, facilitates this identification. By relying on established databases, the method can quickly and efficiently identify relevant protein interactions, building on pre-existing knowledge.

[0053] By doing so, the method identifies the interactions involved in at least one metabolic pathway linked to a relevant biological process, allowing to understand the functional implications of the proteomic data.

[0054] Finally, the step of generating data groups from the identified interactions efficiently organizes the information, making the results more accessible and understandable. By grouping the data according to interactions and metabolic pathways, the proteomic analysis report provides a structured view of the active protein networks in the sample.

[0055] Advantageously, said database comprises a plurality of protein groups, or clusters, the proteins of the same group participating in the same biological process, the same metabolic pathway or the same interaction network. Where appropriate, the identification step comprises an enrichment sub-step comprising, for each protein present in each sample, the identification in said database of all the protein groups containing said protein.

[0056] Several databases may be used for the enrichment sub-step, including the Gene Ontology (GO) databases - biological process category; KEGG Pathway; Reactome; String and Uniprot. The following databases may be excluded for the enrichment sub-step: BioCyc; BioGRID; COG; COMPARTMENTS; DIP; DISEASES; EggNOG; Ensembl; FlyBase; GEO; Gene Ontology - Cellular Component; Gene Ontology - Molecular Function; HPRD; HUGO; Intact ; InterPro; MINT; OMIM; PDB; Pfam; ProteomeHD; PubMed Central; RefSeq; SGD; SIMAP; SMART; SwissModel; FABRICS; WikiPathways; WormBase; proGenomes.

[0057] Advantageously still, said enrichment sub-step comprises, for each protein present in each sample and for each identified group, the calculation of one or more statistical values ​​associated with said group indicating the probability that said protein is actually part of said group. Said statistical value may for example be a p-value and / or a false discovery rate, determined from the number of total proteins making up the group in said database and the number of proteins present in each sample and forming part of said group.

[0058] Preferably, the identification step includes a sub-step of selecting the identified groups according to the biological process, metabolic pathway or interaction network with which they are associated, and / or according to said statistical value with regard to one or more given conditions. For example, it may be possible to: a. exclude groups associated with biological processes and metabolic pathways of the genomic, phylogenetic, molecular function, cellular compartment, localization, disease and medical, sequence, yeast, bacteria type; b. keep only the groups for which the false discovery rate is less than 0.05 and whose p-value is less than or equal to 0.05, so as to keep only the significant groups; c. keep only the groups for which the number of total proteins composing the group in said database is greater than 250, so as to filter out processes and pathways that are too general.

[0059] Other thresholds than those mentioned may be considered, and these thresholds may be selected in advance by a user. One may consider using only one of the conditions mentioned or combining all of these conditions. Other conditions may also be considered, particularly those based on the number of proteins present in each sample and forming part of the said group.

[0060] Advantageously, in a grouping sub-step of the generation step, the identified and, where appropriate, selected groups are grouped with said data to form said data group.

[0061] If desired, the identification step includes, prior to the enrichment sub-step, a matching sub-step including, for each protein present in each sample, searching in a database, for example STRING, for a correspondence between the unique identifier of said protein specific to mass spectrometry and an identifier of said protein in said database comprising a plurality of protein groups. This step thus makes it possible to match the Uniprot codes resulting from the spectrometry and the String codes necessary for implementing the enrichment sub-step.

[0062] Advantageously, the identification step comprises, prior to the enrichment sub-step, a protein filter sub-step among said data to delete the data relating to the proteins present in each sample whose absolute value of a “fold change” type value is less than a given threshold, for example 0.5, unless the p-value of this value is less than a given threshold, for example 0.05. This ensures that the enrichment sub-step is implemented only for proteins whose expression, in the first sample, has varied significantly compared to the second sample, unless the change is significant.

[0063] Advantageously, it may be provided that the generation step includes, subsequent to the grouping sub-step, a statistical analysis sub-step of said data groups to estimate, for each protein of the data groups, a p-value representing the significance of the variation in the expression of this protein, in the first sample compared to the second sample. Where appropriate, said comparative value will be calculated only for the data groups for which said p-value is less than or equal to a threshold value, for example 0.05. Said statistical analysis may be an analysis of variance (ANOVA) and / or an analysis by statistical test, in particular a Tukey-Kramer test.

[0064] In one embodiment of the invention, the generated proteomic analysis report comprises at least one other graph indicating a plurality of biological processes each associated with one of the data groups and comprising, for each biological process indicated and for each protein present in each biological sample and involved in this biological process, a graphic element indicating said value relating to the expression level of this protein, the graphic elements being linked together to form a network whose nodes are these graphic elements and the lines indicate the interactions of the proteins represented by these graphic elements with each other.

[0065] Advantageously, the presentation of the proteomic analysis report in the form of a graph indicating a plurality of biological processes provides an intuitive and understandable visualization of protein interactions and their involvement in various biological processes. This graphical representation facilitates the understanding of complex relationships between proteins and allows for the rapid identification of relevant metabolic pathways and biological processes.

[0066] The addition of a graphical element indicating the relative value of each protein's expression rate for each biological process enhances the richness of the information provided. By directly visualizing the expression variations for each protein within a process, it becomes possible to identify key proteins that may be regulated or affected in a given context.

[0067] Creating a network where graphical elements are linked together to indicate protein interactions is particularly advantageous. This network provides an overview of protein relationships, allowing for an understanding of how proteins interact and cooperate within specific metabolic pathways. The nodes and lines of the network provide a clear visual representation of protein interactions, facilitating the identification of central proteins or hubs that may play a crucial role in a biological process.

[0068] In doing so, this graphical and network approach to the proteomic analysis report ensures a combination of quantitative and relational information, allowing for clear analysis and interpretation of proteomic data.

[0069] Alternatively, the generated proteomic analysis report comprises at least one further graph indicating a plurality of biological processes each associated with one of the data groups and comprising a graphic element for each data group, this graphic element indicating the biological process associated with said data group. The graphic elements are linked together to form a network whose nodes are these graphic elements and the lines indicate the relationships between the biological processes represented by the nodes they link.

[0070] In one embodiment of the invention, the comparative value of the protein expression levels of a data group is calculated from the average of the ratios between the values ​​relating to the protein expression levels of said group in the first sample and the values ​​relating to the protein expression levels of said group in the second sample. It will be noted that in the case where the expression levels are formulated in such a way logarithmic, said comparative value may be calculated by the difference between the average of the values ​​relating to the expression rates of the proteins of said group in the first sample and the average of the values ​​relating to the expression rates of the proteins of said group in the second sample.

[0071] If desired, the comparative value of protein expression rates in a data group can be weighted by a weight determined from a p-value of said data group.

[0072] Advantageously, calculating the comparative value of protein expression levels in a data set from the average of the ratios provides a standardized and robust method for assessing differences in protein expression between samples. By using the average of the ratios between the relative values ​​of protein expression levels in the first sample and the second sample, the method ensures that each protein contributes equally to the overall comparative value, thus avoiding potential biases that could result from extremely abundant or rare proteins.

[0073] By doing this, it is possible to minimize the impact of extreme variations or outliers, as the average of the ratios tends to be less sensitive to individual fluctuations than other measures. This makes the comparative value more representative of the entire data set, thus providing a more balanced view of protein expression differences between samples.

[0074] Additionally, based on the average ratios, the method facilitates comparisons between different data groups or studies. It allows for quick and easy comparisons of comparative values ​​from different groups or samples, as they are assured that each value has been calculated in the same standardized manner. This facilitates decision-making and interpretation of results, especially in large-scale studies or meta-analyses.

[0075] In the context of this invention, proteomic analysis often requires a comparative assessment of protein expression levels between different samples. A commonly used method to quantify these differences is the calculation of ratio abundances. This method involves taking the average of the base-2 logarithms (Log2) of the expression level ratios. This logarithmic transformation allows for a symmetrical distribution of the ratios, thus facilitating subsequent statistical analyses.

[0076] When multiple samples have undergone the same treatment and there are multiple control samples, a more refined approach is adopted. In this case, the average of the Log2 ratios is calculated by taking the average of the expression levels of the samples treated samples and dividing it by the average expression levels of the control samples. This method provides better precision because it accounts for intra-group variation and allows for a more robust comparison between treated and control samples.

[0077] One variation considered in the context of this invention is the use of pooled abundances. Rather than focusing on ratios, this method is based on the average expression levels of samples of the same type. This approach can be particularly useful when intra-group variations are small and the objective is to obtain an overview of general trends in expression levels.

[0078] In a cumulative embodiment of the invention, the comparative value of the protein expression levels of a data group is calculated by multiplying said average of the ratios by the ratio between the number of proteins in said data group and the total number of proteins present in each biological sample.

[0079] Advantageously, the comparative value of the protein expression rates of a data group may be obtained by multiplying the average of the ratios by the ratio between the number of proteins in the data group and the total number of proteins present in each biological sample introduces a weighting dimension. This weighting makes it possible to take into account the relative representativeness of a group of proteins compared to the entire sample. Alternatively, the comparative value of the protein expression rates of a data group may be calculated by multiplying said average of the ratios by the ratio between the number of proteins in said data group and the total number of proteins that may be functionally involved in the biological process(es) associated with the data group.

[0080] By incorporating this weighting, the method ensures that larger or more representative protein groups have a proportionally greater impact on the comparative value. This is particularly advantageous for preventing smaller or less representative protein groups from exerting a disproportionate influence on the results. This ensures that the comparative value more accurately reflects the relative importance of each protein group in the overall sample.

[0081] Additionally, this calculation method provides standardization that facilitates comparison between different samples or studies. By considering both the average expression ratios and the representativeness of the protein group, the method provides a standardized metric that can be used to consistently compare protein expression differences between different samples or contexts.

[0082] In summary, this weighted approach improves the robustness and relevance of the comparative value, allowing a more precise and representative proteomic analysis of the biological samples studied.

[0083] In one embodiment of the invention, the step of calculating at least one comparative value of the expression levels comprises a sub-step of selecting the data groups from said calculated comparative values, and in that the first graph of the generated proteomic analysis report indicates only the biological processes associated with the selected data groups. Advantageously, the introduction of the sub-step of selecting the data groups from the calculated comparative values ​​allows a focus on the most relevant information. By specifically selecting the data groups based on their comparative values, the method ensures that only the most significant or relevant protein expression variations are highlighted. This avoids overloading the analysis report with information that might be less relevant or less significant for the current study.It should be noted that, in the case where the generation step includes a statistical analysis sub-step, said selection sub-step is not necessary, although it is conceivable.

[0084] The fact that the first graph of the generated proteomic analysis report only shows the biological processes associated with the selected data sets reinforces this focus. By presenting only the most relevant biological processes, the report becomes more readable and easier to interpret. Thus, it is possible to focus on the metabolic pathways and / or biological processes that are most affected or predominant between the compared samples.

[0085] This selective approach also improves reporting efficiency by reducing the time and effort required to analyze and interpret data. By focusing on the most relevant information, the process facilitates rapid and informed decision-making, which is particularly advantageous in time-sensitive settings, such as clinical or agricultural diagnostics or high-throughput studies.

[0086] In a cumulative embodiment of the invention, the sub-step of selecting the data groups from said calculated comparative values ​​comprises at least one statistical analysis of said calculated comparative values.

[0087] Advantageously, the integration of a statistical analysis of the calculated comparative values ​​in the data group selection sub-step strengthens the rigor and reliability of the selection process. By using statistical methods, the process ensures that the Selected data sets are not only relevant, but also statistically significant. This minimizes the risk of including protein expression variations that could be due to chance or minor variations.

[0088] Statistical analysis also provides an objective method for assessing the significance of protein expression differences between samples. By relying on statistical criteria, the process avoids potential biases or subjective interpretations that could influence the selection of data groups. This ensures that the proteomic analysis report accurately reflects the true biological differences between samples.

[0089] In addition, the use of statistical analyses allows for greater standardization of the process, so that data groups are selected according to uniform and reproducible criteria, thus facilitating comparison between different reports or studies. This standardization is essential to ensure the consistency and reliability of results, especially in large-scale studies or meta-analyses.

[0090] In summary, integrating statistical analysis into the data group selection process improves the accuracy, reliability, and reproducibility of the proteomic analysis report, providing a solid basis for interpretation and decision-making.

[0091] Alternatively, instead of limiting oneself to a single statistical analysis method, several statistical techniques could be employed simultaneously or sequentially to evaluate the calculated comparative values. For example, nonparametric tests could be used in addition to parametric methods to ensure that the results are not biased by non-normal distributions of the data.

[0092] Additionally, machine learning or artificial intelligence methods could be integrated to complement traditional statistical analysis. These methods could help identify complex patterns or trends in data that might not be immediately evident with traditional statistical techniques.

[0093] As an alternative to simple comparative value analysis, an assessment of the variance or dispersion of the data can also be performed to identify groups of data that exhibit particularly high or low variability, thus providing additional insights into the stability or reliability of protein expression measurements.

[0094] Finally, to strengthen the robustness of the selection, a cross-validation system can be implemented. By dividing the data into training and test sets, and repeating the analysis on several subsets, it will be possible to ensure that the data groups selected are truly representative and are not the result of artifacts or bias in a particular subset of data.

[0095] In one embodiment of the invention, the sub-step of selecting the data groups from said calculated comparative values ​​comprises a variance analysis.

[0096] Advantageously, integrating analysis of variance (ANOVA) into the data group selection sub-step provides a robust and proven method for assessing differences between calculated comparative values. ANOVA is specifically designed to compare means across multiple groups, making it particularly suitable for assessing protein expression variations between different biological samples.

[0097] The use of ANOVA allows us to determine whether the observed differences between data groups are statistically significant. This provides additional assurance that the identified variations are not simply the result of random variation or experimental error, but reflect true biological differences between samples.

[0098] Additionally, ANOVA provides a measure of effect size, allowing for the assessment of the magnitude of differences between groups. This is particularly useful for determining whether a difference, although statistically significant, is also biologically relevant.

[0099] Additionally, ANOVA is able to handle situations where multiple factors or variables may influence protein expression. This allows for a more nuanced and detailed analysis of the data, taking into account multiple sources of variation.

[0100] In sum, integrating analysis of variance into the data group selection process enhances the rigor and precision of the analysis, ensuring that the results are both statistically valid and biologically relevant.

[0101] Alternatively, instead of using a standard analysis of variance, a multivariate analysis of variance (MANOVA) could be used, allowing the effect of several independent variables on several dependent variables to be assessed simultaneously. This would be particularly useful if one wishes to study the impact of several factors or conditions on a range of proteins or metabolic pathways.

[0102] Additionally, principal component analysis (PCA) or factor analysis techniques could be integrated to reduce the dimensionality of the data and identify the main sources of variation in biological samples. These methods could help to visualize relationships between samples and identify the proteins or groups of proteins that contribute most to these variations.

[0103] As an alternative to ANOVA, nonparametric tests, such as the Kruskal-Wallis test, could be used, particularly if the data do not meet the assumptions of normality or homogeneity of variances required for ANOVA. These tests provide additional flexibility for analyzing data that do not follow a normal distribution.

[0104] Finally, to enhance the robustness of the analysis, statistical inference techniques, including bootstrap or permutation techniques, could be applied to estimate the distribution of the tested statistics and provide more precise confidence intervals for the estimates. These methods address potential problems related to small samples or non-standard distributions, thus ensuring a more reliable and robust analysis.

[0105] In an alternative or cumulative embodiment, the sub-step of selecting the data groups from said calculated comparative values ​​comprises an analysis by statistical test, in particular a Tukey-Kramer test.

[0106] Advantageously, the integration of a statistical test analysis, and more specifically a Tukey-Kramer test, in the data group selection sub-step provides a rigorous method for comparing the means of several groups. The Tukey-Kramer test is particularly suitable for multiple comparisons, allowing the identification of groups that differ significantly from each other after an analysis of variance.

[0107] Using the Tukey-Kramer test ensures that the differences identified between data groups are truly significant, while controlling the Type I error rate, which is the risk of falsely concluding a difference. This is particularly beneficial when comparing many groups or samples, as it minimizes the risk of false positive errors.

[0108] In addition, the Tukey-Kramer test provides confidence intervals for differences between means, thus providing a measure of the precision and reliability of the estimates. These intervals allow us to determine not only whether differences exist, but also the magnitude of those differences.

[0109] In summary, the adoption of the Tukey-Kramer test in the data group selection process enhances the rigor and accuracy of the analysis. It ensures that the selected data groups reflect true biological differences between samples, while providing detailed information on the magnitude and reliability of these differences.

[0110] In one embodiment of the invention, said first graph comprises: a. for each biological process indicated on this graph, at least one first graphic element graduated and associated with this biological process; b. for each treatment applied to at least one biological sample from which data was received at the reception step, at least one second graphic element graduated and associated with this treatment; and c.for each group of data containing the data relating to the proteins present in a biological sample having undergone a treatment associated with a second graphic element and involved in the same biological process associated with a first graphic element, a third graphic element linking this first graphic element to this second graphic element, the width of the third graphic element at the level of the first and second graphic elements indicating said comparative value obtained for the group of data in relation to the graduation of said first and second graphic elements.

[0111] Alternatively, it may be provided that said width of the third graphic element is calculated from a combination of the number of total proteins in the protein group associated with said data group, the number of proteins present in each sample and forming part of the protein group associated with said data group, said comparative value, and the p-value associated with said data group.

[0112] Advantageously, the introduction of graduated graphical elements associated with each biological process and each applied treatment provides a clear and intuitive visualization of the data. These graphical elements allow users to quickly identify and compare protein expression levels associated with different biological processes and treatments.

[0113] The third graphical element, connecting the first element associated with the biological process to the second element associated with the treatment, provides a direct visual representation of the relationships between treatments and biological processes. The width of the third graphical element at the level of the first and second graphical elements, indicating the comparative value, allows for a rapid interpretation of the magnitude of the effect of a treatment on a specific biological process. This visual representation facilitates understanding of the complex interactions between treatments and biological processes, without having to navigate through large data tables.

[0114] Additionally, the graduation of the first and second graphical elements provides a reference scale, allowing users to quickly quantify protein expression variations. This standardized graduation facilitates comparison between different reports or studies, ensuring consistent and uniform interpretation of results.

[0115] In summary, the combination of these graphical elements in the first graph significantly improves the readability and usefulness of the proteomic analysis report. It allows for rapid, accurate, and intuitive interpretation of the data, thus facilitating informed decision-making in research or clinical settings.

[0116] Advantageously, the structuring of the graphic elements by forming a substantially complete circle, with the biological processes arranged on a first semicircle and the treatments on a second semicircle, offers a harmonious and intuitive visualization of the interactions between the treatments and the biological processes. This circular arrangement facilitates direct comparison between the different biological samples and the associated treatments, while maximizing the use of the space available on the graph.

[0117] The third graphical elements, forming ropes connecting biological processes to treatments, provide a direct visual representation of the relationships between the two. This configuration allows users to quickly identify the specific effects of each treatment on different biological processes, without having to navigate lists or tables. The width of these ropes, indicating the comparative value, adds an additional dimension of information, allowing for rapid interpretation of the magnitude of a treatment's effect.

[0118] The semicircle arrangement also facilitates comparison between biological samples. For example, by placing the first biological sample that underwent a first treatment near the control sample and the third biological sample that underwent a second treatment, users can quickly assess the relative effects of each treatment compared to the control.

[0119] In short, this circular configuration, combined with the visual representation of the relationships between treatments and biological processes, makes the proteomic analysis report both aesthetically pleasing and informative, facilitating the understanding and interpretation of complex data.

[0120] Advantageously, the introduction of a step of identifying claims supported by one or more biological processes enriches the informative scope of the proteomic analysis report. By associating specific claims with biological processes, users can quickly understand the importance or clinical relevance of the observed proteomic variations.

[0121] The successive arrangement of the first graphic elements associated with biological processes supporting the same claim on the first graphic facilitates the visualization and understanding of the relationships between biological processes and their clinical or functional implications. This organization allows for intuitive reading and direct comparison of biological processes related to a specific claim.

[0122] The fourth graphic element, superimposed on the first set of graphic elements associated with biological processes supporting a particular claim, provides a clear visual representation of the claim itself. This allows users to quickly identify and understand the clinical or functional implications of proteomic variations without having to refer to external explanatory texts.

[0123] Therefore, the addition of claims and their direct visual representation on the graph reinforces the clinical and functional relevance of the proteomic analysis report. This makes it possible to easily and quickly identify the importance of the observed proteomic variations and to make informed decisions based on clearly presented and organized information.

[0124] Advantageously, the integration of another graph representing a comparative statistical distribution of the expression rate values ​​provides a detailed and in-depth perspective of proteomic variations within a specific biological process. This representation allows users to visualize not only the average variations, but also the complete distribution of expression rates for each protein involved.

[0125] This comparative statistical distribution provides a more nuanced understanding of the data. It can identify specific trends, anomalies, or patterns that might not be immediately obvious when examining averages or comparative values ​​alone. For example, it might reveal subgroups of proteins that respond similarly or differently to a treatment, or highlight specific proteins that stand out for their behavior.

[0126] Furthermore, by providing a detailed view of the distribution of expression levels for all proteins involved in a biological process, the proteomic analysis report becomes a more robust tool for decision-making. It is then possible to use this information to identify potential targets for therapeutic interventions, or to better understand the underlying mechanisms of a disease or response to treatment.

[0127] In summary, the addition of a comparative statistical distribution graph enriches the proteomic analysis report, providing a more comprehensive and detailed perspective of proteomic variations, and thus strengthening users' ability to interpret and act on the data presented.

[0128] Of course, the various features, variants and embodiments of the invention may be combined with each other in various combinations to the extent that they are not incompatible or mutually exclusive. Description of Figures

[0129] In addition, various other characteristics of the invention emerge from the appended description given with reference to the drawings which illustrate non-limiting embodiments of the invention and where:

[0130] [Fig. 1] is a detailed flow diagram representing the different steps of a method for generating a proteomic analysis report for the comparison of two biological samples according to a first embodiment of the invention, highlighting the interactions, filtering, grouping and statistical analyses involved in the process.

[0131] [Fig. 2] is a schematic representation of the samples used in the proteomic analysis. It shows three triplets for each condition of the proteomic analysis, CA and CB, as well as three control triplets and three vehicles. Each triplet is clearly identified by labels, allowing easy distinction between the different sample types.

[0132] [Fig. 3] is a schematic representation of a structured database used to store and manage information about proteins, their unique identifiers, their interactions, and associated metabolic pathways. This database is essential for the analysis and comparison of biological samples.

[0133] [Fig. 4] is a graph illustrating the relationships between proteins present in biological samples. Each vertex in the graph represents a protein, identified by a unique label. The shape of the vertices varies depending on the result of a statistical test, and their size is proportional to the comparative value of the expression levels. The edges of the graph indicate the proteins involved in the same biological process, thus providing an overview of protein interactions.

[0134] [Fig. 5] is a detailed graphic representation of the biological processes and associated treatments. It shows a semicircle composed of seven biological processes, each associated with a first graduated graphic element. A second semicircle completes the first, representing the treatments applied to the samples. Graphic elements connect these two semicircles, indicating the relationships between the biological processes and the treatments. These elements are superimposed by fourth graphic elements indicating the claims associated with each process.

[0135] [Fig. 6] represents, schematically and partially, an example of implementation of a step of grouping data into data groups.

[0136] [Fig. 7] represents, schematically and partially, a method for generating a proteomic analysis report for the comparison of two biological samples according to a second embodiment of the invention.

[0137] We have represented in [Fig. 1] a process for generating a proteomic analysis report implemented by computer for the comparison of two biological samples XI and X2.

[0138] The method begins with a step E0 of implementing mass spectrometry on biological samples XI and X2. This step subsequently allows the detection of peptides present in each biological sample, the identification of proteins from the detected peptides, and the association of each identified protein with a unique identifier.

[0139] Step E0 is followed by a data reception step E1. During this step, data relating to a plurality of proteins present in each biological sample are received. This data includes an identifier for each protein and a value relating to the expression level of this protein in each sample. For each identified protein, the data also includes information on the peptides detected and composing this protein, and in particular their number.

[0140] After step E1, a detection and identification step E2 is implemented. During this step, data from mass spectrometry are used to detect the peptides present in each biological sample. These detected peptides then serve as markers to identify the corresponding proteins. Each protein is identified by associating the detected peptides with a database containing a list of proteins, each associated with a unique identifier. Once the proteins have been identified, the abundance ratios are calculated. These ratios represent the quantitative relationship between the proteins present in the two biological samples compared. If several samples with the same treatment and If several control samples are present, the average of the Iog2 of the ratios of the average expression rates of the first to the average expression rates of the second is calculated. This value, representative of the expression rates, makes it possible to obtain a standardized measure of the variation in protein expression between the samples. It will be possible to provide for the comparative values ​​to be associated with statistical values, such as p-values ​​or false discovery rates FDR.

[0141] Step E2 is followed by a filtering step E3. Before clustering, protein data that only contains one peptide are removed.

[0142] Next, step E3 leads to step E4 of grouping the data into data groups. To do this, a specialized database is used. In this database, each protein is associated with a unique identifier, thus ensuring traceability and a clear distinction between proteins.

[0143] In addition, the database also includes protein interactions. For each protein listed, the database provides information on the other proteins with which it interacts within the cell as well as the metabolic pathways to which the proteins belong.

[0144] In addition, the database provides information on the metabolic pathways themselves, including the biological processes they regulate or participate in. Thus, when grouping proteins, said proteins are classified according to their direct interactions and according to the metabolic pathways and biological processes with which they are associated; so as to group proteins into coherent data groups, reflecting the biological processes taking place in the sample studied.

[0145] It may be envisaged to use more than one database for the implementation of step E4 without departing from the scope of the present invention. For example, it may be possible to use a database making it possible to obtain the interactions of the proteins from the identified proteins and then a second database making it possible to obtain metabolic pathways from the identified interactions. It may also be envisaged to use different databases, each making it possible to obtain metabolic pathways and then to consolidate the results from these different databases.

[0146] Furthermore, step E4 may be carried out automatically, the grouping being carried out by one or more data grouping algorithms, or alternatively semi-automatically, at least part of the grouping then being carried out under the supervision of a human being. Tl

[0147] Step E4 is followed by a calculation step E5. For each data group, a comparative value of the protein expression rates is calculated based on the values ​​relative to the expression rate. This comparative value is obtained by calculating the average of the 1000 log2 of the ratios between the values ​​relative to the protein expression rates of the group in the first sample and the values ​​relative to the protein expression rates of the group in the second sample. This average is then multiplied by the ratio between the number of proteins in the data group and the total number of proteins present in each biological sample.

[0148] Thus, the comparative value is given by the following expression: [Math.

[0149] where: X is the comparative value for a given data group, n is the number of proteins in the data group, N is the total number of proteins potentially functionally involved in the biological process(es) associated with the data group, TElj is the expression rate of protein i in the first sample and TE2j is the expression rate of protein i in the second sample.

[0150] Alternatively, the comparative value could be obtained by calculating the average of the Iog2 of the values ​​relative to the expression levels of the group proteins in the samples.

[0151] It may be provided that the comparative value of the protein expression rates of a data group is weighted by a weight determined from a p-value of said data group.

[0152] Step E5 leads to a step E6 of identifying claims. The claims supported by certain biological processes are identified, and the graphic elements associated with these processes are arranged successively on the graph. For each identified claim, a fourth graphic element indicating said claim is superimposed on all of the first graphic elements associated with biological processes supporting this claim.

[0153] Finally, step E6 is followed by step E7 of data group selection. This selection is based on the calculated comparative values, and statistical analyses, such as analysis of variance or the Tukey-Kramer test, can be used for this selection.

[0154] Step E7 is followed by a step E8 of generating a proteomic analysis report. A report is generated comprising at least a first graph indicating the biological processes and the associated graphic elements. For certain processes, another graph may be generated representing a comparative statistical distribution of the values ​​relating to the expression rate. It may be provided that a proteomic analysis report template is stored in a computer memory and that step E8 includes automatic filling of said model using the various data calculated during the previous steps.

[0155] Step E9 of the method consists of saving the generated proteomic analysis report. This saving can be done in digital form on a local or remote storage medium. In parallel with the saving, the report is transmitted to a remote digital terminal via communication networks. This transmission can be carried out by e-mail, file transfer or any other data communication method. In addition, the report can also be sent to a printing unit to be printed on a paper medium. Said printing can be carried out via a locally or remotely connected printer.

[0156] We have represented in [Fig. 2] a diagram illustrating a set of samples used for proteomic analysis. The set is organized into four distinct groups, each dedicated to a specific condition or sample type.

[0157] Samples for testing the CA proteomic analysis condition are represented by three triplets, labeled CAI, CA2, and CA3, respectively. Each triplet is represented by an ellipse, and the three ellipses are aligned horizontally. They are clearly identified by their label.

[0158] Similar to the CA samples, the samples intended to test the CB proteomic analysis condition are labeled CB1, CB2, and CB3. These triplets are also represented by ellipses, aligned horizontally.

[0159] The control samples, which serve as a reference for the analysis and have not undergone any specific treatment, are labeled NI, N2 and N3. These triplets are represented by triangles, aligned horizontally.

[0160] Finally, the samples that underwent neutral treatment, i.e., they were exposed to a vehicle without an active agent, are labeled M1, M2, and M3. These triplets are represented by rectangles, aligned horizontally. The arrangement of the samples in [Fig. 2] is such that each group is clearly separated from the others, allowing for rapid identification and easy visual comparison. Next to each group, a detailed legend provides information on the sample type, the treatment undergone (if any), and other relevant details from the claims, such as the analytical methods used and the unique identifiers associated with each protein in the samples.

[0161] We have represented in [Fig. 3] the detailed structure of a database used in the process of generating a proteomic analysis report for the comparison of two biological samples, in particular during the implementation of step E4.

[0162] The database is designed to store and manage information related to a variety of proteins present in biological samples. Each entry in the database corresponds to a specific protein and is associated with a unique identifier, thus ensuring the uniqueness and traceability of each protein.

[0163] In addition to the unique identifier, each entry contains data related to the protein expression level in biological samples. This data includes relative values ​​that reflect the abundance or concentration of the protein in each sample.

[0164] The database also includes information on the metabolic pathways associated with each protein. These metabolic pathways are associated with specific biological processes. Thus, for each protein, the database provides details on the biological process(es) in which it is involved.

[0165] Additionally, the database contains information on protein interactions. Each listed protein is associated with at least one other protein in the database, indicating an interaction between them. These interactions are crucial for understanding metabolic pathways and protein networks within biological samples.

[0166] Additionally, the database is structured to facilitate the grouping of data into data groups. These groups are formed based on metabolic pathways and biological processes. Thus, proteins involved in the same biological process or metabolic pathway are grouped together, allowing for a coherent and structured analysis.

[0167] We have represented in [Fig. 4] a graph illustrating the relationship and characteristics of the proteins present in samples XI and X2 received at step E1. This graph can be incorporated into the report generated at step E8.

[0168] Each vertex in the graph represents a distinct protein from the set of samples. These vertices are clearly identified by a label, which corresponds to the unique identifier assigned to each protein during the mass spectrometry step. The shape of each vertex is determined based on the result of a Tukey-Kramer statistical test. If the comparative value comparison test is deemed significant for a protein, the vertex is representative takes the form of a disc. Otherwise, where the test is not significant, the vertex is represented by a square.

[0169] The size of these disks and squares is proportional to the comparative value of the protein they represent. The higher this value, the larger the size of the peak, thus reflecting the relative importance of the protein in the comparison between samples XI and X2.

[0170] The color opacity of each vertex is adjusted to indicate the intensity of protein expression. Higher opacity means more intense protein expression, while lower opacity indicates less expression.

[0171] The edges of the graph, on the other hand, establish connections between proteins that are involved in the same biological process. These edges allow you to quickly and efficiently visualize the proteins that interact or cooperate in a given biological process.

[0172] In addition, vertex clusters are formed to group proteins with a similar function or interacting with each other. These clusters are clearly delineated and provide an overview of the main functions represented by proteins from samples XI and X2.

[0173] We have represented in [Fig. 5] a graph generated during step E8, this graph illustrating the relationship between proteins and associated biological processes, in the context of the comparison of two biological samples XI and X2.

[0174] The graph is structured into two main semicircles. The first upper semicircle has seven distinct vertices P1, P2, P3, P4, P5, P6, and P7, each formed by an arc symbolizing a specific biological process identified in step E4 and associated with a data group, after selection in step E6. These biological processes are those in which the proteins in the samples are involved. Each vertex is marked with a label, providing a clear identification of the biological process it represents. Each vertex has a tick mark.

[0175] The second lower semicircle, on the other hand, highlights the treatments Cl and C2 applied to the biological samples. Each treatment is also represented by a vertex, in the form of a graduated arc of a circle and each of these vertices Cl and C2 is connected to the vertices PI to P7 of the first semicircle by ropes. These ropes symbolize the specific proteins that are present in the samples that have undergone a particular treatment and that are involved in a given biological process. The width of these ropes at the vertices indicates the comparative value of proteins for the associated treatment and biological process, as determined in step E5, thus reflecting the graduation of the graphic elements. These graduated graphic elements thus provide a visual representation of the protein expression levels associated with each biological process and treatment.

[0176] Alternatively, it may be provided that the width of each chord is calculated from a combination of the number of proteins theoretically functionally involved in the biological process(es) associated with the data group, the number of proteins present in each sample and involved in the biological process(es) associated with said data group, said comparative value, and the p-value associated with said data group. Within the first semicircle, the vertices representing the biological processes are arranged so that those supporting the same claim A1, A2, A3, identified during step E7, are positioned successively. Three distinct claims are thus highlighted, each grouping together a set of biological processes. For each identified claim, a fourth graphic element is superimposed on all the vertices associated with biological processes supporting this claim.This fourth graphic element indicates the claim in question, providing a clear overview of the claims and the biological processes associated with them.

[0177] [Fig. 6] shows an example of the implementation of step E4 of grouping data into data groups that can be used in the method of [Fig. 1],

[0178] Step E4 includes a first step E41 of identifying the interactions of the proteins present in each biological sample with each other. This step E41 is implemented using one or more databases each comprising a plurality of groups of proteins, or clusters, the proteins of the same group participating in the same biological process, the same metabolic pathway or the same interaction network.

[0179] In a first sub-step E411, the data from step E3 are filtered to remove the data relating to the proteins present in each sample whose absolute value of the expression rate, or any other value likely to indicate a change in expression such as a “fold change” type value, is lower than a given threshold, for example 0.5, unless the p-value of this value is lower than a given threshold, for example 0.05.

[0180] In a second sub-step E412, for each protein present in each sample, a match is sought in a database, for example STRING, between the unique Uniprot identifier of said protein specific to mass spectrometry and a code STRING of said protein allowing this protein to be identified in the database(s) necessary for identifying interactions.

[0181] In a third enrichment sub-step E413, for each protein present in each sample, all the protein groups containing said protein, or clusters, are identified in the database(s). In this sub-step, with regard to the fields of analysis concerned by the present invention, it may be envisaged to use several databases, and in particular to use the Gene Ontology (GO) databases - biological process category; KEGG Pathway; Reactome; String and Uniprot. In particular, all of the following databases may be excluded for the enrichment sub-step: BioCyc; BioGRID; COG; COMPARTMENTS; DIP; DISEASES; EggNOG; Ensembl; FlyBase; GEO; Gene Ontology - Cellular Component; Gene Ontology - Molecular Function; HPRD; HUGO; Intact; InterPro; MINT; OMIM; PDB; Pfam; ProteomeHD; PubMed Central; RefSeq; SGD; SIMAP; SMART; SwissModel; FABRICS; WikiPathways; WormBase; proGenomes.

[0182] Furthermore, in sub-step E413, for each protein present in each sample and for each identified group, a p-value and an FDR are calculated from the number of total proteins making up the group in said database and the number of proteins present in each sample and forming part of said group.

[0183] Optionally, step E41 may include a sub-step E414 for selecting the identified groups according to the biological process, metabolic pathway or interaction network with which they are associated, and / or according to said statistical value with regard to one or more given conditions. For example, it may be possible to: a. exclude groups associated with biological processes and metabolic pathways of the genomic, phylogenetic, molecular function, cellular compartment, location, disease and medical, sequence, yeast, bacteria type; and b. keep only the groups for which the false discovery rate is less than 0.05 and whose p-value is less than or equal to 0.05, so as to keep only the significant groups; and c. keep only the groups for which the number of total proteins making up the group in said database is greater than 250, so as to filter out processes and pathways that are too general.

[0184] It will be noted that, in the case where step E414 is implemented, step E7 of selecting the method of [Fig. 1] can be deleted.

[0185] Step E4 also includes a second step E42 of generating said data groups from said identified interactions.

[0186] In a first sub-step E421, the identified and, where appropriate, selected groups are grouped with said data from step E3 to form said data groups.

[0187] Furthermore, in a statistical analysis sub-step E422 of said data groups, for each protein of the data groups, a p-value representing the significance of the variation in the expression of this protein, in the first sample compared to the second sample, is calculated. It will thus be noted that, in step E5, said comparative value can be calculated only for the data groups for which said p-value is less than or equal to a threshold value, for example 0.05.

[0188] [Fig. 7] shows a method for generating a proteomic analysis report implemented by computer for the comparison of two biological samples XI and X2 according to a second embodiment of the invention.

[0189] Similar to the method of [Fig. 1], the method comprises steps E0, E1 and E2 of implementing mass spectrometry on the biological samples XI and X2, receiving the data, detecting and identifying the peptides and calculating the different values ​​relating to the expression levels.

[0190] The method also comprises the filtering step E3 of the method of [Fig. 1]. However, the method further comprises another filtering step E31, in which only the data relating to the proteins present in the first biological sample XI and whose expression level is significantly increased compared to the second sample X2 are deleted from said data. Said filter may for example be implemented by comparing the abundance ratio of each protein between the first sample and the second sample to a given threshold value, predetermined or set in advance by a user, and by comparing the p-value associated with this protein to another given threshold value.

[0191] At the end of steps E3 and E31, the method comprises a step E3' for generating missing data. In this step E3', in the case where the method is implemented for a group of first samples having undergone the same treatment, in particular a triplicate, and one of said values ​​relating to the expression rate of one of the proteins is missing for one of the samples of said group, an estimate of said value may be calculated from the existing values, relating to the expression rate of this protein in the other samples of said group.

[0192] By way of example, step E3' may comprise, for each protein of which one of said values ​​is missing in said data: a sub-step of normalization of the existing values ​​in said data and relating to the expression level of this protein in the samples of said group; and a sub-step of random selection of a value according to a statistical distribution, in particular normal, corresponding to an interval of less than 2.5% of the distribution of the normalized values.

[0193] Following the estimation of the missing values, step E3' thus comprises the calculation of the ratios of expression rates between each sample of the same group of samples having undergone the same treatment and between each sample of the same group of control samples; averages of these expression rates over these groups of samples, ratios of these averages or any other value of the "fold change" type likely to reflect the change in the expression of the protein under the influence of a condition compared to a control sample. Step E3' may also include a sub-step of calculating a statistical value associated with this estimation of missing value, such as a p-value or a false discovery rate or FDR.

[0194] At the end of step E3', in a similar manner to the method of [Fig.l], the method comprises a step E4 of grouping the data into data groups, a step E5 of calculating the comparative values ​​of the protein expression levels, a step E6 of identifying claims, an optional step E7 of selecting the data groups, a step E8 of generating a proteomic analysis report and a step E9 of saving the generated proteomic analysis report.

[0195] It should be noted, however, that in step E8 of generating the report, a network different from that of [Fig. 4] is generated.

[0196] In this embodiment, the network is a graph indicating a plurality of biological processes each associated with one of the data groups and comprising a graphical element for each data group, this graphical element indicating the biological process associated with said data group.

[0197] It should be noted that in these figures the structural and / or functional elements common to the different variants may have the same references.

[0198] Of course, various other modifications may be made to the invention within the scope of the appended claims.

Claims

Claims

1. A method for generating a computer-implemented proteomic analysis report for comparing two biological samples (ECH1) and (ECH2), the first biological sample (XI) of which has undergone treatment and the second biological sample (X2) being a control sample, characterized in that it comprises the following steps: a. a step (El) of receiving data relating to a plurality of proteins present in each biological sample, comprising: for each protein of said plurality of proteins, an identifier of said protein and at least one value relating to the expression level of this protein in each sample; b.a step (E2) of grouping the data into data groups using at least one database containing a list of proteins, each associated with an identifier and a metabolic pathway linked to a biological process; each data group containing the data relating to the proteins present in each biological sample and involved in the same biological process; c. for each data group, a step (E3) of calculating, from the values relating to the expression rate, at least one comparative value of the expression rates of the proteins of this group in the first sample compared to the expression rates of the proteins of this group in the control sample; d.a step (E4) of generating a proteomic analysis report comprising at least one first graph indicating at least one biological process associated with a group of data and, for each biological process indicated on this graph, at least one graphic element indicating said comparative value obtained for the group of data associated with this biological process.

2. Method for generating a proteomic analysis report for the comparison of two biological samples according to the preceding claim, characterized in that it comprises a prior step of implementing mass spectrometry on the two biological samples to generate said data relating to a plurality of proteins present in each biological sample.

3. Method for generating a proteomic analysis report for the comparison of two biological samples according to the preceding claim, characterized in that each protein listed in the database is associated with a unique identifier, and in that the prior step of implementing mass spectrometry includes the detection of peptides present in each biological sample, the identification of proteins present in each biological sample from the identified peptides and the association with each identified protein of one of said unique identifiers.

4. Method for generating a proteomic analysis report for the comparison of two biological samples according to one of the preceding claims, characterized in that the step (E2) of grouping the data into data groups comprises a step of identifying, using at least one database containing a list of proteins each associated with an identifier and at least one interaction with another protein in this list, said interaction being involved in at least one metabolic pathway linked to a biological process, the interactions of the proteins present in each biological sample with each other, and a step of generating said data groups from said identified interactions.

5. Method for generating a proteomic analysis report for the comparison of two biological samples according to one of the preceding claims, characterized in that said comparative value of the expression levels of the proteins of a group of data is calculated from the average of the ratios between the values relating to the expression levels of the proteins of said group in the first sample and the values relating to the expression levels of the proteins of said group in the second sample.

6. Method for generating a proteomic analysis report for the comparison of two biological samples according to one of the preceding claims, characterized in that step (E3) comprises a sub-step (E3.1) of selecting the data groups from said calculated comparative values, and in that the first graph of the generated proteomic analysis report indicates only the biological processes associated with the selected data groups.

7. Method for generating a proteomic analysis report for the comparison of two biological samples according to one of the preceding claims, wherein said first graph comprises: a. for each biological process indicated on this graph, at least one first graphic element graduated and associated with this biological process; b. for each treatment applied to at least one biological sample from which data was received in the reception step, at least one second graphic element graduated and associated with this treatment; and c. and, for each group of data containing the data relating to the proteins present in a biological sample having undergone a treatment associated with a second graphic element and involved in the same associated biological process to a first graphical element, a third graphical element connecting this first graphical element to this second graphical element, the width of the third graphical element at the level of the first and second graphical elements indicating said comparative value obtained for the group of data in relation to the graduation of said first and second graphical elements.

8. Method for generating a proteomic analysis report for the comparison of two biological samples according to the preceding claim, the method being implemented for a plurality of biological samples including a first biological sample (XI) having undergone a first treatment, a second control biological sample (X2) and at least one third biological sample having undergone a second treatment, characterized in that the set of first graduated graphic elements associated with the biological processes involving proteins present in the first, second and third biological samples substantially form a first semicircle, in that the set of second graduated graphic elements associated with the first and second treatments substantially form a second semicircle completing the first semicircle,and in that the third graphic elements form cords each connecting a first graphic element of the first semicircle to a second graphic element of the second semicircle.,

9. Method for generating a proteomic analysis report for the comparison of two biological samples according to one of claims 7 or 8, the method comprising a step of identifying at least one claim supported by one or more biological processes involving proteins present in each biological sample, in that the first graphic elements associated with biological processes supporting the same claim are arranged successively on the first graphic, and in that said first graphic comprises, for each identified claim, a fourth graphic element indicating said claim and superimposed on all of the first graphic elements associated with biological processes supporting this claim.

10. Method for generating a proteomic analysis report for the comparison of two biological samples according to one of the preceding claims, characterized in that the generated proteomic analysis report comprises, for at least one biological process associated with a group of data, at least one other graph representing a comparative statistical distribution of the values relating to the expression rate of all the proteins present in each biological sample and involved in this biological process.

Images