Keywords:
pharmacogenomics
drug
cancer
cancer-drug
drug-target
transcriptomics
gene network
bipartite network
Poster:
Alberto Berral-Gonzalez, Bioinformatics and Functional Genomics Group. Cancer Research Center (CiC-IBMCC, CSIC/USAL/IBSAL), Spain
Monica M. Arroyo, Department of Chemistry. Universidad Pontificia Catolica de Puerto Rico (PCUPR), Puerto Rico
Santiago Bueno-Fortes, Bioinformatics and Functional Genomics Group. Cancer Research Center (CiC-IBMCC, CSIC/USAL/IBSAL), Spain
Diego Alonso-Lopez, Bioinformatics and Functional Genomics Group. Cancer Research Center (CiC-IBMCC, CSIC/USAL/IBSAL), Spain
Javier De Las Rivas, Bioinformatics and Functional Genomics Group. Cancer Research Center (CiC-IBMCC, CSIC/USAL/IBSAL), Spain
Short Abstract:
Pharmacogenomics is a growing field that studies the use and effect of drugs together with the genomic information of individuals. The study of the genome-wide activity of the genes (i.e. the expression of the genes or their genetic variation) may be related to differences in the effects of drugs. This information, integrated into clinical trials and drug development, could help to understand the behaviour and/or results of the action of these drugs in complex diseases. Cancer is a complex disease with hundreds of clinically approved drugs. Pharmacogenomics allows better mapping of the targets of cancer drugs and potential interacting secondary agents, but there are many drugs whose mechanisms of action have not been fully deciphered. The study of these drug-targets can lead to possible new treatments or an improvement of existing ones. This study comprised a large-scale screening method to find associations of many chemical substances and human genes using transcriptomic profiling. It’s focused on a compilation of two types of drugs: approved by the Food and Drug Administration (FDA) and not approved by the FDA (No-FDA). The second one includes more than two thousand chemical compounds. These compounds are related to the transcriptomic profiles of 60 human cell lines, for which gene expression profiles are also available. The standard expression of each gene versus the standardized activity of each biological compound was used to calculate pairwise correlations for all available gene-drug combinations. With these data, global bipartite networks were built to further study the interactions between compounds and their targets, to better unravel their mechanisms of action. In addition, all these data were included in GEDA, an online web-tool that allows the user to navigate the networks obtained. Furthermore the user can access the used data and review the results obtained in this work. The results obtained provide information on the complex action of the studied compounds, presenting a relational and integrative view to address the different biomolecular effects that each drug can produce.
Keywords:
cancer mutations
breast cancer
metastasis
exome
whole exome sequencing
gene mutation
polygenic signature
machine learning
Poster:
Fernando Bueno-Gutierrez, Bioinformatics and Functional Genomics Group. Cancer Research Center (CiC-IBMCC, CSIC/USAL/IBSAL), Spain
Oscar Gonzalez-Velasco, Bioinformatics and Functional Genomics Group. Cancer Research Center (CiC-IBMCC, CSIC/USAL/IBSAL), Spain
Santiago Bueno-Fortes, Bioinformatics and Functional Genomics Group. Cancer Research Center (CiC-IBMCC, CSIC/USAL/IBSAL), Spain
Jose M Sanchez-Santos, Department of Statistics. University of Salamanca (USAL), Spain
Javier De Las Rivas, Bioinformatics and Functional Genomics Group. Cancer Research Center (CiC-IBMCC, CSIC/USAL/IBSAL), Spain
Short Abstract:
Whole Exomes Sequences (WES) from more than six thousand Breast Cancer Primary Tumors were used to find a polygenic mutation prognosis signature specific of tumors that are likely to develop metastasis. Samples were divided in two groups: primary tumors that did not develop metastasis in the first five years after biopsy (negative samples: 6334), and primary tumors that, while being metastasis-free at the time of biopsy, they developed metastasis in that time frame (positive metastatic samples: 382). A train set was used to find mutations exclusive from positives, and these were ranked based on frequency and pathogenicity-scores. Then, on a balanced validation set, supervised k-means-clustering was used to exclude the 20% samples from this set that were most difficult to separate. Clustering was based on the mutations selected on the train set. On the remaining validation samples, step-wise regression was carried to output a refined list of the mutations selected. The train/validation split was repeated 10,000 times and the 600 most frequent mutations defined the final signature. With this mutation signature, we were able to correctly classify 84% of the samples from a test set consisting of 100 samples from each class.
Presentation 57: In silico analysis and homology modelling of human monocarboxylate transporters involved in cancer
Keywords:
Cancer
Lactate
Monocarboxylate transporters
AR-C155858
Poster:
Andres Patricio Ibacache Chia, Pontificia Universidad Católica de Chile, Chile
Andreas Schüller, Pontificia Universidad Católica de Chile, Chile
Jimena Alejandra Sierralta Jara, Universidad de Chile, Chile
Short Abstract:
Cancer is the second leading cause of death worldwide and corresponds to the uncontrolled development of abnormal cells that infiltrate and destroy normal tissue. To support their proliferation, tumor cells depend on lactate-based metabolism. Lactate is transported over membranes by monocarboxylate transporters (MCTs), which in humans are divided into 14 types. Of these, MCTs 1 and 4 are primarily responsible for the transport of lactate over the plasma membrane rendering them potential targets for the inhibition of tumor development.
Despite the interest in generating MCT antagonistic drugs, few selective inhibitors for these transporters have been developed so far. This is due to the transmembrane nature of these proteins, that complicate the determination of their three-dimensional structures and make it difficult to obtain relevant information about the residues that mediate their transport function. As of now, there are no crystal structures available for MCT 1 and 4, a situation that has also made difficult a more detailed study of this type of transporter.
Here, we present comparative protein structure models of the three-dimensional structures of MCT1 and MCT4. The models were built with the MODELLER software and validated with the ProSa-web and Saves v5.0 servers. The template used for construction was the first structure of a human MCT transporter published recently thanks to an electronic cryo-microscopy technique, a human MCT2 with high sequence identity to MCT 1 and 4 (59% and 46%, respectively). Next, we performed molecular docking of AR-C155858, a commercial MCT1-selective inhibitor. Our results suggest that AR-C155858 binds to residues present in the 7-10 transmembrane segments of MCT1 and not of MCT4, due to the presence of non-conserved residues in the inhibitor binding site of each transporter.
These structural models can provide a starting point for the structural and functional analysis of human MCTs and the design of potential inhibitors of the activity of these proteins.