These results proven the distinct functions and advantages of our method in prioritizing important genes responsible for drug resistance based on time-course RNA-seq data. Open in a separate window Fig 7 Assessment of the effectiveness of the top 5 genes prioritized from the DryNetMC with that by other methods, including DEseq2 and GSNCA.RNA-seq data of U87MG cells was used to test its distance to sensitive DBTRG-05MG cells or resistant LN-18 cells, based on the expression pattern similarity of the determined genes evaluated using pair-wised DTW distance. LN-18 cells did not.(TIF) pcbi.1007435.s001.tif (302K) GUID:?520F1D9B-77A7-4047-BE00-E3A322B60FD9 S2 Fig: Quantifying and comparing monotonic changes or adaptive changes of the TCGs in sensitive and resistant cells. (A-B) Illustration showing the definition of scores for monotonic response and adaptive response. (C) Assessment of monotonic reactions of TCGs between the sensitive cells and the resistant cells. (D) Assessment of adaptive reactions of TCGs between the sensitive cells and the resistant cells. One-tailed Wilcoxon rank sum test p-values were used to assess the statistical significance. These results indicated the TCGs in the resistant cells tended to have higher adaptive response scores but lower monotonic response scores compared to the TCGs in the sensitive cells.(TIF) pcbi.1007435.s002.tif (1.0M) GUID:?CBFD6162-6AC7-497C-97DD-94E8A3A21679 S3 Fig: Illustration and validation of the DryNetMC for network inference based on a simulated dataset. (A) A true network SB590885 with standard motifs, such as positive and negative opinions loops and crosstalk. A system of ODEs, in the form of = 1,2,,5), was built to generate the original time course gene manifestation data. The connection confidents (experimental data were used to validate the drug sensitivity prediction based on the similarity of the temporal patterns of the prioritized important genes. We compared our method with other methods including the standard differential expression analysis and differential co-expression networkCbased method. The computational method developed with this study is generally relevant for the analysis of time-course RNA-seq data designed for studying drug resistance in many cancer types. Materials and methods The computational pipeline for the time-course transcriptome-based modeling and characterization of the GRNs underlying drug resistance is definitely illustrated in Fig 1. Below, we describe the details of each step. Open in Mouse monoclonal to RUNX1 a separate windows Fig 1 The computational method of DryNetMC (differential regulatory network-based modeling and characterization) developed to prioritize key genes responsible for drug resistance.(We) The TCGs were determined as core genes from time-course RNA-seq data of sensitive and resistant cells. (II) The dynamic GRNs for sensitive cells and resistant cells were reconstructed via a approach that incorporates prior info, data interpolation, dynamic systems modeling and regularized regression methods. (III) Subsequently, a differential network was then extracted and its practical enrichment was performed. (IV) Moreover, the features of network topology, local entropy and adaptation dynamics were analyzed to measure the importance of each node in the differential network for prioritizing key genes responsible for drug resistance. (V) In addition, the above node importance measurement was incorporated into a differential regulatory network-based biomarker (DryNB) model for predicting drug response of medical individuals. (VI) Furthermore, experimental data and statistical significance test were used to validate the effectiveness of the key genes prioritized by DryNetMC. Recognition of temporal changes in gene manifestation The RNA-seq data for both sensitive cells and resistant cells were measured at following drug treatment. The natural RNA-seq reads were processed using a standard pipeline [17C20] (observe details in S1 Text). Because gene expressions display temporal changes over time, we designed the following algorithm to globally select significant temporally changing genes (TCGs) by comparing expression levels between any two time points for a given gene. A given gene with manifestation level (= 0, 1, , or for some and and the fold switch of its manifestation level between two time points was greater than and could become chosen according to the percentage or the number of the selected TCGs for further network building. We empirically select the first hundreds of TCGs (~5%-10% genes of the whole transcriptome) for the following network modeling and visualization. As such, thresholds for and may become arranged SB590885 accordingly. In this study, was arranged to 10 and was arranged to 5, as in our earlier study [21]. SB590885 Data interpolation Given that the total number of time points (i.e., (= 0, 1, , 0, 1, , and the derivative of (for example, = 100) points from at time is the number of nodes in the network. The function is the connection strength from to is a constant quantity accounting for the effects of degradation or self-activation. represents the prior info and association between gene and gene in an initial correlation network and denote is definitely sufficiently small (since is chosen large enough as mentioned above). Therefore, the above continuous model (i.e., Eq (2)) can be rewritten mainly because = (= (= (= (= (= (= = (is the error term. are mutually self-employed normal random variables with means 0, 0, ?, 0 and variances represents the Hadamard product of and and and is the penalty weight. Ten-fold cross validation was performed to select the optimal value of that minimizes the mean cross-validated errors. The regression.
