HK1232312B - Systems and methods for predicting a smoking status of an individual - Google Patents

Description

System and method for predicting smoking status of an individual

Citation of Related Applications

This application claims priority under 35 U.S.C. §119 to U.S. Provisional Patent Application No. 61/916,443, filed on December 16, 2013, entitled “Systems and Methods for Predicting a Smoking Status of an Individual,” which is incorporated herein in its entirety.

Background Art

Whole-genome microarrays are being used as a practical means to measure genome-wide expression levels and gain biological insights into a variety of conditions. This approach has also been used to assess the body's response to exposure to active substances and predict the resulting phenotype. Molecular alterations in the transcriptome of large airway cells from smokers in response to cigarette smoke exposure can be detected even in the absence of overt histological abnormalities. This observation suggests that transcriptomic data may be useful for assessing the response of biological systems to exposure to a variety of substances.

In many product risk assessment studies, obtaining samples from the desired primary site (e.g., the airway) is invasive and inconvenient. As an alternative, peripheral blood sampling is minimally invasive and widely used in the general population. Therefore, there is a focus on identifying and establishing biomarkers that can be reliably used in peripheral blood, which serves as a surrogate tissue.

Early attempts to explore molecular biomarkers focused on identifying differentially expressed genes between case and control groups. The latest methods are dedicated to predicting new cases more and more, thereby promoting enhanced diagnosis, improved prognosis and promoting personalized medicine. However, the development of robust and universal computational methods for clinical applications remains challenging. Regarding smoking-related diseases, diagnostic signatures in peripheral blood samples have been identified. At least two studies have shown that differentially expressed genes can distinguish subjects with early non-small cell lung cancer from control subjects or subjects with non-malignant lung diseases (Rotunno, M., Hu, N., Su, H., Wang, C., Goldstein, A.M., Bergen, A.W., Consonni, D., Pesatori, A.C., Bertazzi, P.A., Wacholder, S. et al. (2011). A gene expression signature from peripheral whole blood for stage I lung adenocarcinoma. Cancer Prev Res) (Phila) 4, 1599-1608; Showe, M.K., Vachani, A., Kossenkov, A.V., Yousef, M., Nichols, C., Nikonova, E.V., Chang, C., Kucharczuk, J., Tran, B., Wakeam, E., et al. (2009). Gene expression profiles in peripheral blood mononuclear cells can distinguish patients with non-small cell lung cancer from patients with nonmalignant lung disease. Cancer Res 69, 9202-9210).

Summary of the Invention

A computational system and method are provided for identifying a robust blood-based gene signature that can be used to predict an individual's smoker status. The gene signature described herein can distinguish between current smokers and never-smokers or former smokers, thereby enabling accurate prediction of an individual's smoker status.

In certain aspects, the systems and methods of the present disclosure provide a computerized method for evaluating a sample obtained from a subject. The computerized method includes receiving, via a receiving circuit, a data set associated with the sample, the data set comprising quantitative expression data for LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17. A processor generates a score based on the received data set, the score indicating a predicted smoking status of the subject. The predicted smoking status can classify the subject as a current smoker or a non-current smoker.

In some implementations, the dataset further includes quantitative expression data of IGJ, RRM2, SERPING1, FUCA1, and ID3. In some implementations, the score is the result of a classification scheme applied to the dataset, wherein the classification scheme is determined based on the quantitative expression data in the dataset.

In certain implementations, the method further comprises calculating a fold change value for each of LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17, and determining that each fold change value satisfies at least one criterion. The criterion may require that for at least two independent population data sets, each respective calculated fold change value exceeds a predetermined threshold.

In certain aspects, the systems and methods of the present disclosure provide a computerized method for assessing a sample obtained from a subject. The apparatus includes a device for detecting the expression levels of genes in a gene signature in a test sample, the gene signature including LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17. The apparatus also includes a device for correlating the expression levels with a classification of a smoker's status, and a device for outputting the classification result of the smoker's status as a prediction of the subject's smoker's status.

In certain aspects, the systems and methods of the present disclosure provide a kit for predicting an individual's smoker status. The kit comprises: a set of reagents for detecting the expression levels of genes in a gene signature in a test sample, the gene signature comprising LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17; and instructions for using the kit to predict an individual's smoker status.

In certain aspects, the systems and methods of the present disclosure provide a kit for assessing the effect of an alternative to a smoking product on an individual. The kit comprises: a set of reagents for detecting the expression levels of genes in a gene signature in a test sample, the gene signature comprising LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17; and instructions for using the kit to assess the effect of the alternative on an individual. The alternative to a smoking product may be a heated tobacco product (HTP), and the effect of the alternative on the individual may be classifying the individual as a non-smoker.

In certain aspects, the systems and methods of the present disclosure provide a method for assessing a sample obtained from a subject. The method includes receiving a data set associated with the sample via a receiving circuit. The data set includes quantitative expression data for at least five markers selected from a group consisting of: LRRN3, CDKN1C, PALLD, SASH1, RGL1, TNFRSF17, IGJ, RRM2, SERPING1, FUCA1, and ID3. The method further includes generating, via a processor, a score based on the received data set, the score indicating a predicted smoking status of the subject. The predicted smoking status of the subject can classify the subject as a current smoker or a non-current smoker.

The score can be the result of a classification scheme applied to the dataset, wherein the classification scheme is determined based on quantitative expression data in the dataset. The method can further include calculating a fold change value for each of LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17, and determining that each fold change value satisfies at least one criterion. The criterion can require that for at least two independent population datasets, each of the calculated fold change values exceeds a predetermined threshold.

In certain aspects, the systems and methods of the present disclosure provide an apparatus for assessing a sample obtained from a subject. The apparatus includes means for detecting expression levels of genes in a gene signature in a test sample, the gene signature including at least five markers selected from the group consisting of: LRRN3, CDKN1C, PALLD, SASH1, RGL1, TNFRSF17, IGJ, RRM2, SERPING1, FUCA1, and ID3. The apparatus further includes means for correlating the expression levels with a classification of a smoker's status, and means for outputting the classification result of the smoker's status as a prediction of the subject's smoker's status.

In certain aspects, the systems and methods of the present disclosure provide a kit for predicting an individual's smoker status. The kit comprises: a set of reagents for detecting expression levels of genes in a gene signature in a test sample, the gene signature comprising at least five markers selected from the group consisting of LRRN3, CDKN1C, PALLD, SASH1, RGL1, TNFRSF17, IGJ, RRM2, SERPING1, FUCA1, and ID3; and instructions for using the kit to predict an individual's smoker status.

In certain aspects, the systems and methods of the present disclosure provide a kit for assessing the effect of an alternative to a smoking product on an individual. The kit comprises: a set of reagents for detecting the expression levels of genes in a gene signature in a test sample, the gene signature comprising at least five markers selected from the group consisting of: LRRN3, CDKN1C, PALLD, SASH1, RGL1, TNFRSF17, IGJ, RRM2, SERPING1, FUCA1, and ID3; and instructions for using the kit to assess the effect of the alternative on an individual. The alternative to a smoking product can be a high-pressure smoking product (HTP), and the effect of the alternative on the individual can be classifying the individual as a non-smoker.

BRIEF DESCRIPTION OF THE DRAWINGS

Further features of the present disclosure, its nature and various advantages will become apparent upon consideration of the following detailed description taken in conjunction with the accompanying drawings, in which like reference characters refer to like parts throughout, and in which:

FIG1 is a flow chart of a method for identifying a set of genes and obtaining a classification model based on the set of genes.

2 is a flow chart of a method for assessing a sample obtained from a subject.

3 is a block diagram of an exemplary computing device that may be used to implement any component of any of the computerized systems described herein.

4A , 4B and 4C are volcano plots of differentially expressed genes in the sample datasets.

5A , 5B, 5C, 5D, 5E, and 5F are various box plots indicating the classification schemes of different studies.

DETAILED DESCRIPTION

This article describes a computational system and method for identifying a robust blood-based gene signature that can be used to predict an individual's smoker status. Specifically, the gene signature described herein can distinguish between subjects who are currently smoking and subjects who have never smoked or have quit smoking.

As used herein, a "robust" gene signature is a gene signature that maintains strong performance across research, laboratories, sample sources, and other population factors. Importantly, a robust label should be detectable, even if it is detectable in a group of population data that includes large individual differences. In order to avoid overly optimistic reporting of the performance of the label, the robustness in the entire data set should also be properly verified.

One goal of the present disclosure is to obtain a gene signature that can accurately predict an individual's smoker status. In order to evaluate the performance of the gene signature, the prediction results are shown in the table below, which shows the predicted status in each row and the true status in each column. Table 1 below shows an example of a way to demonstrate the prediction results. The first row of the table indicates the true current smoker and non-current smoker populations that the predicted sample should be associated with the current smoker, and the second row of the table indicates the true current smoker and non-current smoker populations that the predicted sample should be associated with the non-current smoker.

Table 1

A perfect predictor is one that accurately predicts all current smokers as current smokers (true positives would be 100% and false negatives would be 0%), and accurately predicts all non-current smokers as non-current smokers (true negatives would be 100% and false positives would be 0%). As described herein, individuals are classified according to smoking status (e.g., current smoker, non-current smoker, former smoker, never smoker, etc.), but in general, one of ordinary skill in the art will understand that the systems and methods described herein are applicable to any classification scheme.

In order to evaluate the strength of a predictor, various metrics based on the values in the prediction results table can be used. In this article, one metric is called "sensitivity," which is the ratio of individuals in the current smoker group who are accurately classified as current smokers. In other words, the sensitivity metric is equal to the number of true positives divided by the sum of true positives and false negatives, or TP/(TP+FN). A sensitivity value of 1 represents a perfect classification of current smokers. In this article, another metric is called "specificity," which is the ratio of individuals in the non-current smoker group who are accurately classified as non-current smokers. In other words, the specificity metric is equal to the number of true negatives divided by the sum of true negatives and false positives, or TN/(TN+FP). A specificity value of 1 represents a perfect classification of non-current smokers. In order to be considered a strong predictor, it is desirable that the sensitivity and specificity values are high. Although sensitivity and specificity metrics are used herein to evaluate the performance of a predictor, in general, any other metric, such as the predictive value (TP/(TP+FP)) of a positive test or the predictive value (TN/(TN+FN)) of a negative test, can also be used without departing from the scope of this disclosure.

The system and method described herein constructs a predictive model by the following steps: first, genes with high fold changes in expression levels are identified from different training data sets. Then, the identified set of genes is verified using an independent data set. After verification, the set of genes is tested by evaluating the blood transcriptome of a subject whose smoker status is known and comparing the expression levels of the identified set of genes for individuals with one smoker status with those with another smoker status. The resulting set of genes that have been successfully verified and tested is referred to as a "gene signature" in this article.

Genetic tags can be used to accurately classify individuals into specific predicted smoker status groups. In addition, by being able to accurately predict the smoker status of an individual, the genetic tags can detect the use of various HTPs by comparing the results of individuals using HTPs with the results of individuals smoking conventional cigarettes. Genetic tags can be used in situations where compliance with smoking behavior is needed. In one example, the predicted smoker status of an individual (as determined by the genetic tags) can be used in clinical trials of HTPs to identify whether biological changes occur in an individual after the individual switches to the HTP or when biological changes occur in the individual. In general, genetic tags can be used in any health-related research on monitoring smoking, quitting smoking or switching to an HTP.

In one example, data from several publicly available gene expression data sets were obtained to analyze blood samples of current smokers and non-smokers or former smokers. Preselecting genes from each independent study based on high fold change genes is advantageous because it enhances the robustness of the label in different studies and ensures that the prediction model will not be biased by a single data set. Validated with an independent data set, the independent data set is derived from a clinical study designed to explore novel biomarkers for COPD. In addition, according to another clinical study, the blood transcriptome of smokers who switched from conventional cigarettes (which burn tobacco) to HTP (which does not burn tobacco; referred to herein as Tobacco Heating System (THS) 2.1) for 5 consecutive days was evaluated and compared with smokers who continued to smoke conventional cigarettes. The label described herein performs very well in classifying current smokers and non-current smokers, as demonstrated by the performance of the label using an independent data set. In addition, the effect of switching to THS 2.1 for 5 consecutive days can be detected in the blood transcriptome because subjects who switched to THS 2.1 were classified as non-current smokers. This suggests that the gene signatures and systems and methods described herein can be used not only to determine smoker status but also to assess the short-term effects of smoking.

Using signatures based on a limited number of genes is advantageous in terms of reducing cost and workload compared to using the entire transcriptome, because analysis is ultimately based on quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) measurements. With qRT-PCR, the investment in equipment and running costs (e.g., reagents) is more favorable than with microarrays.

In one example, in a first step, different training data sets are obtained to identify gene signatures. Specifically, two training data sets are used herein: BLD-SMK-01 and QASMC. However, in general, any combination of any number of training data sets can be used without departing from the scope of this disclosure.

About BLD-SMK-01, blood samples collected using the PAXgene blood DNA kit (Qiagen) were obtained from a repository (BioServe Biotechnologies Ltd, Beltsville, MD 20705 USA). At the time of sampling, the subjects were between 23 and 65 years old. Subjects with no history of disease and subjects taking prescription medication were excluded. Current smokers have been smoking at least 10 cigarettes a day for at least 3 years. Former smokers had quit smoking for at least 2 years before sampling and had smoked at least 10 cigarettes a day for at least 3 years before quitting. Current smokers and non-smokers were matched by age and gender. A total of 31 blood samples were obtained from current smokers, 30 blood samples were obtained from never smokers, and 30 blood samples were obtained from former smokers.

Blood samples were also obtained from a Queen Ann Street Medical Center (QASMC) clinical study, which was conducted at the Heart and Lung Centre in London, England, according to Good Clinical Practice (GCP) and registered on ClinicalTrials.gov with identifier NCT01780298. The QASMC study was designed to identify biomarkers or a panel of biomarkers that can distinguish subjects with COPD (current smokers with a smoking history of ≥10 pack-years and GOLD stage 1 or 2) from matched non-smoking subjects (never smokers, former smokers, and current smokers) in three groups of controls. In each of the four groups, samples were obtained from sixty subjects (240 subjects in total). Male and female subjects aged between 40 and 70 years were included. All subjects were matched to the COPD subjects recruited in the study by race, sex, and age (within 5 years of age). Blood samples were sent to AROS Applied Biotechnology AS (Aarhus, Denmark) where they were further processed and then hybridized to Affymetrix Human Genome U133 Plus 2.0 GeneChips as described below.

Total RNA (including microRNA) was isolated using the PAXgene blood miRNA kit (Cat. No. 763134; Qiagen) according to the manufacturer's instructions. The concentration and purity of the RNA samples were determined by measuring the absorbance at 230, 260, and 280 nm using a UV spectrophotometer (NanoDrop ND1000; Thermo Fisher Scientific, Waltham, MA, USA). The integrity of the RNA was further checked using an Agilent 2100 bioanalyzer. Only RNA with an RNA integrity number (RIN) greater than 6 was processed for further analysis.

RNA Preparation and Affymetrix Hybridization. Affymetrix probe sets targeting the 3' end of transcripts were prepared from 50 ng of RNA using NuGEN ^™ Ovation ^™ Whole Blood Reagent and the NuGEN ^™ Ovation ^™ RNA Amplification System V2. The amount of cDNA was measured using a Nanodrop 1000 or 8000 spectrophotometer (Thermo Fisher Scientific) or a SpectraMax 384 Plus (Molecular Devices). The quality of the cDNA was determined by assessing the size of the unfragmented cDNA using an Agilent 2100 Bioanalyzer. The size distribution of the final fragmented and biotinylated products was also monitored using electropherograms. After labeling the cDNA, the fragments were hybridized to GeneChip Human Genome U133 Plus 2.0 arrays according to the manufacturer's instructions. Samples used for target preparation were fully randomized for use on Affymetrix gene expression microarrays.

Taqman qRT-PCR analysis. Reverse transcription was performed with 500 ng of starting RNA using the iScript ^™ cDNA Synthesis Kit (Cat. No. 170-8890; Bio-Rad, Hercules CA, USA) according to the manufacturer's instructions. The cDNA was then accurately diluted to 10 ng/μL. A commercial human universal RNA (UHR) reference (Cat. No. 740000, Agilent Technologies, Santa Clara, CA, USA) was added to the samples as a calibrator to reliably compare data from multiple experiments and instruments. The probes used in the Taqman analysis spanned exons, and five housekeeping genes (B2M, GAPDH, FARP1, A4GALT, GINS2) were selected for the data normalization step. The qPCR step was performed using the Assay and Rapid Advanced Master Mix (Cat. No. 444963). In brief, cDNA was diluted to allow 1.25 ng to be coated per well in a 384-well plate. In parallel, a master mix (containing Taqman assay reagents and Taqman advanced mix) was prepared for each Taqman assay. The final reaction volume was 10 μL. qPCR was run using a Viia7 instrument (Life Technologies) and automatic baseline and default C _t threshold settings were applied to analyze the results. When a Universal Human Reference (UHR) sample was added, the C _t value for each gene was normalized (by subtraction) relative to the UHR C _t value and then normalized relative to the GAPDH housekeeping gene value (generating so-called ΔΔC _t values).

Taqman primers were obtained from Life Technologies, CA, USA. Table 2 below lists the primer sequences used to perform qRT-PCR.

Table 2

Microarray Analysis - Data Quality Check and Normalization. After investigation of the chip images to detect artifacts on the chip scans, the data were processed through a standard quality control pipeline. Briefly, the raw data files were read using the ReadAffy function of the affy package (Gautier, L., Cope, L., Bolstad, B.M. and Irizarry, R.A. (2004). affy---analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 20, 307-315) [6], which is from the Bioconductor suite of microarray analysis tools (Gentleman, R.C., Carey, V.J., Bates, D.M., Bolstad, B., Dettling, M., Dudoit, S., Ellis, B., Gautier, L., Ge, Y., Gentry, J. et al. (2004). Bioconductor: open software development for computational biology and bioinformatics. The microarray analysis tools were used in the R statistical environment (R Development Core Team (2007). R: A Language and Environment for Statistical Computing). Quality control was performed by generating and inspecting the following: RNA degradation curves (AffyRNAdeg function of the affy package), [09:42:29] normalized unscaled standard error curves, relative logarithmic expression curves (affyPLM package (Brettschneider, J., Collins, F. and Bolstad, B.M. (2008). Quality Assessment for Short Oligonucleotide Microarray Data. Technometrics 50, 241-264)), and the mean of the relative logarithmic expression values. In addition, artifact images (residues of the probe level model) were visually inspected to ensure the absence of spatial effects. Arrays falling below a set threshold were excluded from further analysis during quality control checks.

For population level analysis (i.e., mean fold change studies), data were then normalized using GC-Robust Microarray Analysis (GC-RMA). Microarray expression values were generated from all arrays that passed quality control checks using background correction and quantile normalization (Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J., Scherf, U., and Speed, T.P. (2003). Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4, 249-264). For individual signature prediction models, data were normalized using MAS5 (Affymetrix, I. (2002). Statistical algorithms description document. Technical paper).

Statistical Modeling - Population-Level Analysis. For each comparison, an overall linear model was fitted, and raw p-values were generated for each probe set on the expression array based on a moderated t-statistic. The Benjamini-Hochberg False Discovery Rate (FDR) method was used to correct for multiple testing effects due to the large number of genes evaluated.

Statistical Modeling - Individual Sample Predictive Modeling. To achieve robustness in the predictive models, independent gene expression datasets from blood (GSE15289) and PBMC (GSE42057) were obtained and processed from the National Center for Biotechnology Information Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih.gov/gds/?term=GEO). The dataset (GSE15289) from the NOWAC study (Dumeaux, V., Olsen, K.S., Nuel, G., Paulssen, R.H., A.-L., and Lund, E. (2010a). Deciphering normal blood gene expression variation—The NOWAC postgenome study. PLoS genetics 6, e1000873) included whole blood samples from 285 postmenopausal women aged between 48 and 63 years, including 211 never smokers and 74 current smokers. The data set (GSE42057) of Bahr et al. (Bahr, T.M., Hughes, G.J., Armstrong, M., Reisdorph, R., Coldren, C.D., Edwards, M.G., Schnell, C., Kedl, R., LaFlamme, D.J. and Reisdorph, N. (2013). Peripheral Blood Mononuclear Cell Gene Expression in Chronic Obstructive Pulmonary Disease. American journal of respiratory cell and molecular biology) was derived from peripheral blood mononuclear cell (PBMC) samples collected from 36 current smokers (22 of whom had COPD and 14 were healthy) and 100 former smokers (72 of whom had COPD and 28 were healthy). All subjects were non-Hispanic white.

The subject data extracted from GSE15289 and GSE42057 data sets are used to identify the genes that show high average expression changes between smoker samples and never smokers (or former smokers) samples in each data set. Make L ₁ and L ₂ be M (here, M=1000, but in general, M can be any value) the set of the highest fold change genes relative to two independent data sets (GSE15289 and GSE42057). In order to obtain list L ₁ , data set GSE15289 is classified according to smoker status (current smoker and never smoker), and the average gene expression level of each group is obtained. The difference between the current smoker group and the never smoker group in terms of average gene expression level is referred to as fold change in this article, and in set L ₁ , M genes with the highest fold change are included. In a similar manner, but for current smokers and former smokers, list L ₂ is obtained.

Fig. 1 is the flow chart of the method 100 for differentiating one group of genes and obtaining classification model based on this group of genes.Specifically, method 100 comprises the following steps: counter parameter N is initialized to 1 (step 102), by calculating Matthews correlation coefficient (Matthews Correlation Coefficient, MCC (N)) assess the performance of linear discriminant analysis (lineardiscriminantanalysis, LDA) model (step 104), and judge whether counter parameter equals maximum counter value M (decision block 106).If N is less than M, so method 100 advances to step 108 so that N increases progressively and turns back to step 104 to assess the performance of LDA model by calculating next coefficient MCC (N).When N reaches M (decision block 106), assessment produces the N value ( _NMAX ) (step 110) of maximum MCC value, and core gene list is defined as the intersection (step 112) between two groups of genes _L1 [1:N] and _L2 [1:N].After having differentiated core gene list, calculate LDA model (step 114) based on core gene list.

At step 102 , a counter parameter N is initialized to 1. The counter parameter N goes from 1 to a maximum value M and is incremented at step 108 until N reaches M at decision block 106 .

In step 104, the performance of the LDA model is evaluated by calculating the coefficient MCC (N). Specifically, the performance of the LDA model can be evaluated using a 5-fold cross validation (100 times) for _L1 [1:N] ∩ _L2 [1:N], where the cross validation is the intersection of the highest multiple change N in the set _L1 and the highest multiple change N in the set _L2 . The LDA model is evaluated by calculating MCC (N). The MCC metric combines all true/false positive and negative rates and therefore provides a single-valued fair metric. MCC is a performance metric that can be used as a composite material performance score. MCC is a value between -1 and +1 and is essentially a correlation coefficient between a known binary classification and a predicted binary classification. MCC can be calculated using the following equation:

Where TP: True Positive; FP: False Positive; TN: True Negative; FN: False Negative. However, in general, any suitable technique that produces a composite performance metric based on a set of performance metrics can be used to assess the performance of an LDA model. An MCC value of +1 indicates that the model obtains perfect predictions, an MCC value of 0 indicates that the model predictions are almost random, and an MCC value of -1 indicates that the model predictions are completely inaccurate. The advantage of MCC is that it can be easily calculated when the classifier function is encoded in a way that only class predictions are available. In contrast, for area under the curve (AUC) calculations, the classifier function is required to provide a numerical score. However, in general, any metric that accounts for TP, FP, TN, and FN can be used in accordance with the present disclosure.

In order to calculate MCC, the set of classification categories should first be selected. The BLD-SMK-01 data set is obtained from never smokers, former smokers and current smokers. Figures 4 A, 4B and 4C show the volcano plots of the differentially expressed genes in the BLK-SMK-01 sample. Each volcano plot shows the estimated log2 (fold change) versus log10 (adjusted P value). The P value is calculated based on a moderate t-statistic and adjusted by the Benjamin-Hochberg method. Specifically, Figure 4 A compares the gene expression profiles between current smokers and non-smokers, Figure 4 B compares the gene expression profiles between current smokers and former smokers, and Figure 4 C compares the gene expression profiles between former smokers and never smokers. The volcano plot shown in Figure 4C indicates that there were no differential gene expression changes between never smokers and former smokers (i.e., no trend was observed in Figure 4C), but Figures 4A and 4B indicate that many differential gene expression changes were observed between current smokers and never smokers (Figure 4A) and between current smokers and former smokers (Figure 4B).

Therefore, the colony level transcriptomic analysis of BLD-SMK-01 sample shows that there is not differential gene expression change between never smokers and former smokers in whole blood, and therefore, distinguishing former smokers from never smokers based on blood transcriptome will be very challenging.On the contrary, there are a lot of differentially expressed genes (Fig. 4 A and 4B) between current smokers and never smokers and between current smokers and former smokers.Because difference is not observed between never smoker colony and former smoker colony, only two categories are used to evaluate model in step 104: current smoker and non-current smoker.

Specifically, in step 104, the gene set _L1 [1:N] _∩L2 [1:N] corresponds to the intersection of the highest fold change N of the two independent data sets GSE15289 and GSE42057. Cross-validation is performed based on each prediction model of _L1 [1:N] or _L2 [1:N] to assess whether the results of the LDA model can be generalized to the independent data set. In one example, in order to perform a 5-fold cross-validation on the _L1 [1:N] gene set, the _L1 [1:N] set is randomly divided into five subsets: A, B, C, D, and E. Four (A, B, C, and D) subsets are used to train classifiers using the LDA technique, and the fifth subset (E) is used to test the classifiers trained on the other four subsets. The training and testing process is repeated four more times, wherein each of the other subsets (A, B, C, and D) is used as a test subset to test the classifiers trained on the other four subsets.

In general, the criterion of LDA technology is to classify the input vector x describing n features into y classes. The classification is based on a certain function, which is a linear combination of the observed features. The coefficients of the linear combination are estimated based on the training subset of the data. Specifically, in order to use the LDA technology to train a classifier, a linear combination of gene expression levels in the data is identified from four training subsets. The linear combination is referred to as a classifier in this article and defines a boundary between the predicted smoker status and the predicted non-smoker status. The classifier is used to obtain the predicted status of each individual in the test subset. This process is repeated four times in addition, so that the five subsets are each processed as the test subset once. After the five subsets have each been used as the test subset once, a 5-fold cross validation is completed, and the training data observations (with the features in the L ₁ [1:N] ∩ L ₂ [1:N] set) are divided into five new subsets A ', B ', C ', D ' and E ', thereby causing the second 5-fold cross validation.

The examples described herein are the results of 100 5-fold cross validations, but generally speaking, one of ordinary skill in the art will appreciate that any number of k-fold cross validations can be used without departing from the scope of this disclosure. In addition, the examples described herein are the results of the LDA technique, which forms a classifier based on linear combinations of gene expression levels. However, generally speaking, one of ordinary skill in the art will appreciate that any function of gene expression levels can be used to form a classifier, such as a quadratic function, a polynomial function, an exponential function, or any other suitable function that can form a one-dimensional manifold in R^N to define a classifier.

At step 110, after N reaches a maximum number M, a set of M values of the MCC is considered, and the N value corresponding to the maximum value of the MCC is evaluated as N _max = argmax _N (MCC(N)). As shown in FIG1 , the N _max evaluation step is performed after all M values of the MCC have been calculated. However, generally speaking, one of ordinary skill in the art will appreciate that, alternatively, the value MCC(N) calculated at step 104 can be compared to some predetermined threshold before evaluating the next value MCC(N+1). In this case, if the value MCC is found to exceed the predetermined threshold, method 100 can proceed directly to step 110 and assign the value of N _max to the current N value, disregarding the remaining N values = N _max + 1 to M.

In step 112, the core gene list of the signature is defined by the intersection _L1 [1: _Nmax ] _∩L2 [1: _Nmax ], or the set of genes that are in both _L1 [1: _Nmax ] and _L2 [1: _Nmax ]. As described in this example, only two data sets, _L1 and _L2 , are used. However, in general, one of ordinary skill in the art will appreciate that any number of data sets can be used to calculate MCC values and identify the core gene set used to define the gene signature. Specifically, the intersection of m data sets or the union of pairwise intersections can be used.

In step 114, an LDA model is calculated using the core gene list determined in step 112. Specifically, the LDA model calculated based on the core gene list can be calculated by performing 100 5-fold cross validations or any number of n-fold cross validations.

In one example, the statistical modeling method described in steps 102 to 114 was applied to identify a core gene signature comprising the following six genes: LRRN3, SASH1, PALLD, RGL1, TNFRSF17, and CDKN1C. When classifying samples from current smokers and never smokers, the 5-fold cross-validation (100 times) MCC of this model was 0.77 (with a sensitivity score (Se) of 0.91 and a specificity score (Sp) of 0.85). By design of the method, the core genes in the signature were among the high-fold change genes in both the NOWAC (GSE15289) and Bahr et al. (GSE42057) studies, and the prediction was based on all 77 common genes between the two GSE studies, improving the performance of the LDA model (Se = 0.73, Sp = 0.81). Although all six genes LRRN3, SASH1, PALLD, RGL1, TNFRSF17 and CDKN1C are referred to herein as a core gene signature, one of ordinary skill in the art will understand that any combination of the six genes can be used as a core gene signature, such as any combination of three, four or five of the six genes.

In some embodiments, the genes in the signature are expanded to include an extended gene set that includes other genes that are not in the core set but are associated with high specificity scores and high sensitivity scores. Specifically, when studying the predictive model obtained by utilizing each list of high-fold change genes individually, IGJ, RRM2, ID3, SERPING1, and FUCA1 were repeatedly identified as potential candidates in the signature with high specificity and sensitivity. These five genes are also among the high-fold change genes in the blood transcriptome studied by NOWAC (current smokers vs. never smokers) and Bahr et al. (current smokers vs. former smokers) and are used to expand the core gene signature to an extended signature. When current smokers and never smokers were classified, the cross-validated MCC of the model based on the extended signature (LRRN3, SASH1, PALLD, RGL1, TNFRSF17, CDKN1C, IGJ, RRM2, ID3, SERPING1, and FUCA1) was 0.73 (Se=0.88, Sp=0.84). Although all eleven genes LRRN3, SASH1, PALLD, RGL1, TNFRSF17, CDKN1C, IGJ, RRM2, ID3, SERPING1, and FUCA1 are collectively referred to herein as an extended gene signature, one of ordinary skill in the art will appreciate that any combination of the eleven genes can be used as a core gene signature, such as any combination of five, six, seven, eight, nine, or ten of the eleven genes. In addition, the combination can include a combination of three, four, or five of the six genes in the core gene signature and two, three, or four of the five genes in the additional genes in the extended gene signature.

The results of the LDA model calculated in step 114 are compared with the prediction cross-validation results of the model obtained when learning sparse labels from BLD-SMK-01 alone (that is, without using two public datasets GSE15289 and GSE42057). When predicting smokers versus non-smokers, the 5-fold cross-validation performance of this model produces Sp=0.96 and Se=0.93, which is slightly higher than the performance of the model based on core labels and extended labels. Although the cross-validation specificity and sensitivity (Sp=0.88, Se=0.84) of the prediction model derived by the method described herein result in performance slightly lower than the performance (Sp=0.96, Se=0.93) of the model obtained without using an independent dataset, the prediction model derived herein is advantageous because the model is relevant to a wider range of applications. Specifically, the prediction model derived according to the method of the present disclosure is stable, as demonstrated when validating the model, as described in detail with respect to step 116.

In step 116, verify the LDA model calculated in step 114.By using the former smoker group from BLD-SMK-01 research and the blood data set from QASMC research, perform the verification of LDA model.After QASMC transcriptomic samples are carried out quality control, 52 parts of COPD, 58 parts of current smokers, 58 parts of former smokers and 59 parts of never smokers CEL files are available for prediction.In order to assess the predictive performance of core label and extended label, QASMC samples are divided into two groups: current smoker (COPD and health) and non-current smoker (comprising former smoker and never smoker).These two groups allow to assess the robustness of described label with respect to COPD situation.Use the model prediction each centered data set built for core gene label or extended label.

Table 3 shows the prediction results obtained by using the LDA model for independent data sets of various labels. The format of Table 3 follows the format of Table 1, wherein the predicted classification is shown in different rows and the actual classification is shown in different columns. Specifically, the prediction results shown in Table 3 include the prediction results of the following: core gene labels (first three rows), extended gene labels (middle three rows), labels derived only from BLD-SMK-01 samples (second to last row) and labels (last row) of the gene sets described in Beineke et al. (Beineke, P., Fitch, K., Tao, H., Elashoff, M.R., Rosenberg, S., Kraus, W.E. and Wingrove, J.A. (2012). A whole blood gene expression-based signature for smoking status. BMC pharmacogenomics (BMC medical genomics) 5, 58.). As shown in Table 3, both the core and extended signatures resulted in higher sensitivity and specificity scores than the signature derived from the BLD-SMK-01 sample alone and the signature identified by Beineke.

Table 3

The classification performance of the signature against the QASMC study demonstrated that the model was robust regardless of COPD status (Se = 0.9, Sp = 0.9 for the core signature and Se = 0.91, Sp = 0.90 for the extended signature).

In addition, Figures 5A, 5B, 5D, and 5E show individual box plots indicating the classification schemes for different studies. Specifically, Figures 5A and 5B plot box plots of the posterior probability of a sample being classified as a current smoker based on the LDA model for the BLD-SMK-01 study and the QASMC study, respectively. Figures 5D and 5E plot box plots of the predicted scores of the linear discriminant function for the BLD-SMK-01 study and the QASMC study, respectively. Specifically, samples with negative scores were classified as current smokers, and samples with positive scores were classified as non-current smokers.

Also checked the influence of other covariates such as sex and age.BLD-SMK-01 and QASMC study are balanced with respect to sex and age.There is no statistical association between age or sex and smoking status, as indicated by statistical chi-square test (about BLD-SMK-01, χ ² (sex, smoking status) P value=1; and about QASMC, χ ² (sex, smoking status) P value=0.9) and statistical t-test (about BLD-SMK-01, t-test (age vs. smoking status) P value=0.8; and about QASMC, t-test (age vs. smoking status) P value=0.46).

Additionally, in BLD-SMK-01, each gene in the signature was tested for association with sex and age, and none had an ANOVA P-value below 0.05, although the PALLD gene showed a weak sex effect. Previously identified gene signatures have found effects of sex and/or age and established the need to adjust for these factors. Beineke et al. 2012. Specifically, age is a significant covariate in two public datasets (GSE15289 and GSE42057), because smokers are on average older than never or former smokers. Therefore, this covariate was not included in the predictor because it was not statistically associated with smoking status in the BLD-SMK-01 study. However, despite its superior performance, as defined by specificity and sensitivity scores, the gene signatures described herein were generally unrelated to sex or age. This suggests that the core and extended signatures described herein offer an advantage over known gene signatures in that they do not have to adjust for these factors, thereby simplifying the computational approach.

In order to determine whether the label found can be translated into the exposure biomarker based on qRT-PCR, the subset of 20 randomly selected samples (10 current smokers and 10 never smokers) was subjected to qRT-PCR to measure the expression level of the gene in the extended label. Based on the gene in the extended label, the qRT-PCR data training LDA model for normalization was used, and evaluated by 10 folding cross validations (1000 times, selecting 10 foldings was because sample size was small), and specificity 0.85 and sensitivity 0.96 (table 4) were obtained. When the core label was applied with the same operation, specificity 0.62 and lower sensitivity 0.80 (table 4) were obtained.

Table 4

One goal of the present disclosure is to use a core gene signature and an extended gene signature to determine whether the signature can be used to detect the effects of switching to heated tobacco products (HTPs). To facilitate this goal, data were obtained from the REX-EX-01 study. The REX-EX-01 study was an open-label, randomized, controlled, two-arm, parallel-group study that enrolled 42 healthy smokers aged between 23 and 65 years of age, encompassing both sexes. The study was conducted to compare smokers of conventional cigarettes with smokers who had recently switched to an HTP (referred to herein as Tobacco Heating System 2.1 (THS 2.1)) for five consecutive days. The study was conducted according to Good Clinical Practices (GCP) and is registered on ClinicalTrials.gov with identifier NCT01780714. Blood samples were stored in PAXgene tubes and transported to AROS Applied Biotech in Aarhus, Denmark, where they were further processed and then hybridized to the Affymetrix Human Genome U133 Plus 2.0 gene chip.

In order to test whether the gene signature identified herein provides a sensitive and non-invasive tool for assessing the exposure response in clinical trials, the label is applied to THS 2.1 data, it is determined whether the switch to HTP can be detected in the whole blood transcriptome after five days. The hypothesis of this research is that the whole blood transcriptome of the smoker switched to THS 2.1 is more similar to the whole blood transcriptome of the smoker in the past than the similarity with the whole blood transcriptome of the current smoker. Instead of characterizing the gene expression profile (for example, by extracting label from REX-EX-01 research data) of the specific HTP user who switched for five days, it is expected to identify the exposure response label based on transcriptome, and the label can also serve as the indicator of longer-term conversion mode. This is achieved by establishing core gene label and extended gene label, and core gene label and extended gene label can both distinguish current smoker sample and non-current smoker sample.

After quality control was performed on the CEL files of the REX-EX-01 study, 16 and 18 files remained for conventional cigarette smokers and THS 2.1 users, respectively, on the 5th day. Table 5 below shows the prediction results of the REX-EX-01 samples for the core gene signature (first three rows) and the extended gene signature (last three rows). With regard to the extended gene signature, individuals who still smoke conventional cigarettes (current smokers) were mainly classified as current smokers (69%), while subjects who switched to THS 2.1 were mainly classified as non-current smokers (89%). With regard to the core signature, the accuracy rate of current smokers was the same (69%), and 78% of subjects who switched to THS 2.1 were classified as non-current smokers. Therefore, both the core gene signature and the extended gene signature can predict that the samples obtained from HTP users are samples of non-current smokers.

Table 5

The results shown in Table 5 are consistent with the initial hypothesis that the blood transcriptome of subjects switching to THS begins to resemble that of former rather than current smokers, despite the fact that there were no significant differences in nicotine and cotinine exposure between THS 2.1 and conventional cigarettes (data not shown).

In addition, Figure 5C plots a box plot of the posterior probability of a sample being classified as a current smoker based on the REX-EX-01 data from the LDA model, and Figure 5F plots a box plot of the prediction scores from the linear discriminant function based on the REX-EX-01 data. Samples with negative prediction scores are classified as current smokers, while positive prediction scores indicate non-current smoker status.

Compared with the gene label of the measurement that relies on single gene, gene expression spectrum analysis provides the comprehensive and more complete view of the biological process under normal situation and pathological situation.When the expression trend of multiple genes is combined together, it is also possible to derive the label or the classifier for specifying physiological state from the exposure reaction for disease state.Although the tissue that is mainly affected can provide the sample representing the state of normal state, exposure or pathological state more accurately, it is usually impractical to use tissue biopsy to classify the experimenter.Because it is convenient to draw blood using minimally invasive technology, huge prospects are arranged aspect biomarker exploration based on the label of blood.In this research, two groups of biomarkers based on whole blood have been identified, wherein either one can serve as the label of the reaction of health to smoking and therefore can be used as the strong predictor of individual smoking status.

A gene that stood out strongly in this study was LRRN3. In current smokers, expression of LRRN3 was increased compared to non-current smokers. In the REX-EX-01 study, expression decreased significantly between days 0 and 5 in the blood of subjects who switched to HTPs, and remained constant in the blood of subjects who continued to smoke conventional cigarettes. Therefore, LRRN3 appears to be an important gene in both the core and extended signatures for measuring the effects of switching from conventional cigarettes to HTPs. In one example, a gene signature as described includes only LRRN3 and no other genes, or includes LRRN3 and any other genes. Specifically, a gene signature including LRRN3 is able to detect a switch from smoking conventional cigarettes to using HTPs by demonstrating a decrease in LRRN3 expression between days 0 and 5 after the switch.

Systems pharmacology method as herein described allows to construct one or more solid smoker gene signatures based on whole blood that can distinguish current smoker and non-current smoker.Core gene signature as herein described is based on six genes, and extended gene signature is based on core gene signature plus other five genes.Two kinds of gene signatures all have significant accuracy aspect the smoker's status of prediction individuality, as assessed by sensitivity and specificity score.When to the sample application from REX-EX-01 research, described label will use THS 2.1 experimenter after five days to be differentiated as non-current smoker based on whole blood transcriptome data.Therefore, label as herein described provides a kind of sensitive and specific instrument of using minimally invasive sampling assessment exposure response.

FIG2 is a flow chart of a method 200 for assessing a sample obtained from a subject, according to an illustrative embodiment of the present disclosure. Method 200 includes the following steps: receiving a dataset associated with the sample, the dataset comprising quantitative expression data for LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17 (step 202); and generating a score based on the received dataset, wherein the score indicates a predicted smoking status of the subject (step 204). In some embodiments, the dataset received in step 202 further comprises quantitative expression data for IGJ, RRM2, SERPING1, FUCA1, and ID3. In some embodiments, the dataset received in step 202 further comprises quantitative expression data for one or more of CLDND1, MUC1, GOPC, and LEF1.

The score generated at step 204 is the result of applying a classification scheme to the dataset, wherein the classification scheme is determined based on the quantitative expression data in the dataset. Specifically, in the example described herein, a classifier trained on an LDA model can be applied to the dataset received at 202 to determine a predicted classification for the individual.

The gene signatures described herein can be used in a computer-implemented method for assessing a sample obtained from a subject. Specifically, a data set associated with the sample can be obtained, and the data set can include quantitative expression data for LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17 for a core gene signature. A score can be generated based on the received data set, wherein the score indicates the predicted smoking status of the subject. Specifically, the score can be based on a classifier constructed using the LDA model method described herein. The data set can further include quantitative expression data for other markers included in the extended gene signature, IGJ, RRM2, SERPING1, FUCA1, and ID3. The data set can further include quantitative expression data for one or more of CLDND1, MUC1, GOPC, and LEF1.

In some embodiments, the data set includes any subset of any number of the marker set LRRN3, CDKN1C, PALLD, SASH1, RGL1, TNFRSF17, IGJ, RRM2, SERPING1, FUCA1, ID3, CLDND1, MUC1, GOPC, and LEF1. One or more criteria can be applied to the markers included in the signature, such as at least three (or any other suitable number) of LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17, at least two (or any other suitable number) of IGJ, RRM2, SERPING1, FUCA1, and ID3, and at least one (or any other suitable number) of CLDND1, MUC1, GOPC, and LEF1. In general, any signature using a combination of these markers can be used without departing from the scope of the present disclosure.

In some embodiments, the genes in the signature described herein are used to assemble a kit for predicting an individual's smoker status. Specifically, the kit includes a set of reagents for detecting the expression levels of genes in the gene signature in a test sample, and instructions for using the kit to predict an individual's smoker status. The kit can be used to assess the effects of smoking cessation or alternatives to smoking products (such as HTPs) on an individual.

Figure 3 is a block diagram of a computing device for performing any of the methods described herein (such as those described with respect to Figures 1 and 2) or for storing a core gene signature, an extended gene signature, or any other gene signature described herein. Specifically, the gene signature stored on a computer-readable medium includes expression data for LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17. In another example, the gene signature included in the computer-readable medium includes expression data for at least five markers, each selected from the group consisting of LRRN3, CDKN1C, PALLD, SASH1, RGL1, TNFRSF17, IGJ, RRM2, SERPING1, FUCA1, and ID3.

In some specific implementations, components and databases can be implemented in several computing devices 300. Computing device 300 includes at least one communication interface unit, input/output controller 310, system memory, and one or more data storage devices. System memory includes at least one random access memory (RAM 302) and at least one read-only memory (ROM 304). These components are all communicated with a central processing unit (CPU 306) to facilitate the operation of computing device 300. Computing device 300 can be configured in many different ways. For example, computing device 300 can be a conventional stand-alone computer, or alternatively, the functions of computing device 300 can be distributed in multiple computer systems and architectures. Computing device 300 can be configured to perform part or all of modeling, scoring and aggregation operations. In Figure 3, computing device 300 is connected to other servers or systems via a network or local network.

Computing device 300 can be configured to distributed architecture, wherein, database and processor are contained in separate unit or position.Some such units perform main processing functions, and at least comprise general purpose controller or processor and system memory.In such respect, each in these units is attached to communication hub or port (not shown) via communication interface unit 308, and described hub or port are used as the main communication link with other server, client or user computer and other relevant equipment.Communication hub or port itself can have minimum processing power, are mainly used as communication router.Various communication protocols can be the part of system, include but not limited to: Ethernet, SAP, SAS ^TM , ATP, BLUETOOTH ^TM , GSM and TCP/IP.

The CPU 306 includes a processor, such as one or more conventional microprocessors and one or more auxiliary coprocessors, such as a math coprocessor, for offloading the workload from the CPU 306. The CPU 306 communicates with a communication interface unit 308 and an input/output controller 310, and the CPU 306 communicates with other devices, such as other servers, user terminals, or devices, through the communication interface unit 308 and the input/output controller 310. The communication interface unit 308 and the input/output controller 310 may include multiple communication channels for communicating with, for example, other processors, servers, or client terminals simultaneously. Devices that communicate with each other do not need to continuously transmit to each other. Instead, such devices only need to transmit to each other when necessary, and in fact can refrain from exchanging data most of the time, and may need to perform several steps to establish a communication link between the devices.

The CPU 306 also communicates with a data storage device. The data storage device may include a suitable combination of magnetic, optical, or semiconductor memory and may include, for example, RAM 302, ROM 304, a flash drive, an optical disk such as a compact disk, or a hard disk or drive. The CPU 306 and the data storage device may each be, for example, completely located within a single computer or other computing device; or they may be connected to each other by a communication medium such as a USB port, a serial port cable, a coaxial cable, an Ethernet cable, a telephone line, a radio frequency transceiver, or other similar wireless or wired medium, or a combination thereof. For example, the CPU 306 may be connected to the data storage device via a communication interface unit 308. The CPU 306 may be configured to perform one or more specific processing functions.

The data storage device may store, for example: (i) an operating system 312 for the computing device 300; (ii) one or more application programs 314 (e.g., computer program code or computer program product) adapted to direct the CPU 306 according to the systems and methods described herein, and in particular according to the procedures described in detail with respect to the CPU 306; or (iii) a database 316 adapted to store information, which may be used to store information required by the program. In some aspects, the database includes a database storing experimental data and published literature models.

Operating system 312 and application programs 314 may be stored, for example, in a compressed, uncompiled, and encrypted format and may include computer program code. The instructions of the program may be read into the processor's main memory from a computer-readable medium rather than a data storage device (e.g., from ROM 304 or from RAM 302). Although execution of the sequence of instructions in the program causes CPU 306 to perform the process steps described herein, hard-wired circuitry may be used in place of or in combination with software instructions to implement the processes of the present disclosure. Thus, the described systems and methods are not limited to any specific combination of hardware and software.

Suitable computer program code may be provided to perform one or more of the functions described herein. The program may also include program elements such as an operating system 312, a database management system, and "device drivers" that allow the processor to interact with computer peripherals (e.g., a video display, keyboard, computer mouse, etc.) via the input/output controller 310.

As used herein, the term "computer-readable medium" refers to any non-transitory medium that provides or participates in providing instructions to a processor of the computing device 300 (or any other processor of the devices described herein) for execution. Such media can take many forms, including but not limited to non-volatile media and volatile media. Non-volatile media include, for example, optical, magnetic, or magneto-optical disks, or integrated circuit memories such as flash memory. Volatile media include dynamic random access memory (DRAM), which typically constitutes main memory. Common forms of computer-readable media include, for example, floppy disks, diskettes, hard disks, magnetic tape, any other magnetic media, CD-ROMs, DVDs, any other optical media, punch cards, paper tape, any other physical media with a pattern of holes, RAM, PROM, EPROM or EEPROM (Electrically Erasable Programmable Read-Only Memory), FLASH-EEPROM, any other memory chip or cartridge, or any other non-transitory medium from which a computer can read.

Various forms of computer-readable media can be involved in transmitting one or more sequences of one or more instructions to CPU 306 (or any other processor of the device described herein) for execution. For example, the instructions can be initially carried on a disk of a remote computer (not shown). The remote computer can load the instructions into its dynamic memory and send the instructions via an Ethernet connection, a cable line, or even a telephone line using a modem. A communication device local to the computing device 300 (e.g., a server) can receive data on a corresponding communication line and place the data on a system bus for the processor. The system bus transfers the data to the main memory, from which the processor retrieves and executes the instructions. The instructions received by the main memory can optionally be stored in a memory before or after being executed by the processor. In addition, the instructions can be received as electrical signals, electromagnetic signals, or optical signals via a communication port, which are exemplary forms of wireless communication or data streams that carry various types of information.

Each reference cited herein is hereby incorporated by reference in its entirety.

While specific implementations of the present disclosure have been particularly shown and described with reference to specific examples, it will be understood by those skilled in the art that various changes in form and details may be made to these specific implementations without departing from the scope of the present disclosure as defined by the appended claims. The scope of the present disclosure is therefore indicated by the appended claims, and all changes that come within the meaning and range of equivalents of the claims are therefore intended to be embraced therein.

Claims (15)

1. A computer-implemented method for evaluating samples obtained from subjects, comprising: A dataset associated with the sample is received via a receiving circuit, the dataset including quantitative expression data of gene tags containing LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17; and A classifier is applied to the quantitative expression data of the gene tag, and a score is generated by the processor based on the quantitative expression data of the gene tag, wherein the classifier is trained based on a linear discriminant analysis (LDA) model, the LDA model is calculated based on the quantitative expression data of the gene tag, and the score indicates the predicted smoking status of the subject. 2. The computer-implemented method according to claim 1, wherein the quantitative expression data of the gene tag further includes IGJ, RRM2, SERPING1, FUCA1, and ID3. 3. The computer-implemented method according to claim 1 or 2, further comprising calculating the fold change value of each of LRRN3, CDKN1C, PALLD, SASH1, RGL1 and TNFRSF17 in the gene tag. 4. The computer-implemented method of claim 3, further comprising determining that each calculated multiple change value exceeds a predetermined threshold. 5. The computer-implemented method according to claim 1, wherein the LDA model is calculated based on k-fold cross-validation of quantitative expression data of the gene tag. 6. A kit for predicting an individual's smoking status, comprising: A set of reagents for detecting the expression levels of genes in gene tags, including LRRN3, CDKN1C, PALLD, SASH1, RGL1, and TNFRSF17, in test samples; and Instructions for using the kit to predict an individual's smoking status, the instructions including applying a classifier to the detected expression levels of genes in the gene tag, wherein the classifier is trained based on a linear discriminant analysis (LDA) model calculated based on quantitative expression data of the gene tag. 7. The kit according to claim 6, wherein the kit is used to assess the effects of substitutes for smoking products on an individual. 8. The kit according to claim 7, wherein the alternative to the smoking product is a heated tobacco product. 9. The kit according to claim 8, wherein a decrease in LRRN3 expression was detected between 0 and 5 days after the individual began using the heated tobacco product. 10. The kit according to any one of claims 7 to 9, wherein the effect of the substitute on the individual is to classify the individual as a nonsmoker. 11. The kit according to any one of claims 6 to 9, wherein the gene tag further comprises at least one of IGJ, RRM2, SERPING1, FUCA1, and ID3. 12. A kit for predicting an individual's smoking status, comprising: A set of reagents for detecting the expression levels of genes in a gene tag in a test sample, said gene tag comprising up to fifteen biomarkers, wherein at least five biomarkers in said gene tag are selected from the group consisting of: LRRN3, CDKN1C, PALLD, SASH1, RGL1, TNFRSF17, IGJ, RRM2, SERPING1, FUCA1, and ID3; and Instructions for using the kit to predict an individual's smoking status, the instructions including applying a classifier to the detected expression levels of genes in the gene tag, wherein the classifier is trained based on a linear discriminant analysis (LDA) model calculated based on quantitative expression data of the gene tag. 13. The kit of claim 12, wherein the kit is used to assess the effects of a substitute for a smoking product on an individual, wherein the substitute for the smoking product is a heated tobacco product. 14. The kit according to claim 13, wherein a decrease in LRRN3 expression was detected between 0 and 5 days after the individual began using the heated tobacco product. 15. A computer-readable medium comprising computer-readable instructions that, when executed in a computerized system including at least one processor, cause the processor to perform one or more steps of the method according to any one of claims 1 to 5.