With the frenetic growth of high-dimensional datasets in different biomedical domains, there is an urgent need to develop predictive methods able to deal with this complexity. Feature selection is a relevant strategy in machine learning to address this challenge. We introduce a novel feature selection algorithm for linear regression called BOSO (Bilevel Optimization Selector Operator). We conducted a benchmark of BOSO with key algorithms in the literature, finding a superior accuracy for feature selection in high-dimensional datasets. Proof-of-concept of BOSO for predicting drug sensitivity in cancer is presented. A detailed analysis is carried out for methotrexate, a well-studied drug targeting cancer metabolism.

Artificial intelligence (AI) can unveil novel personalized treatments based on drug screening and whole-exome sequencing experiments (WES). However, the concept of "black box" in AI limits the potential of this approach to be translated into the clinical practice. In contrast, explainable AI (XAI) focuses on making AI results understandable to humans. Here, we present a novel XAI method -called multi-dimensional module optimization (MOM)- that associates drug screening with genetic events, while guaranteeing that predictions are interpretable and robust. We applied MOM to an acute myeloid leukemia (AML) cohort of 319 ex-vivo tumor samples with 122 screened drugs and WES. MOM returned a therapeutic strategy based on the FLT3, CBF beta-MYH11, and NRAS status, which predicted AML patient response to Quizartinib, Trametinib, Selumetinib, and Crizotinib. We successfully validated the results in three different large-scale screening experiments. We believe that XAI will help healthcare providers and drug regulators better understand AI medical decisions.

Simple Summary This work shows that the predictions of lethal dependencies (LEDs) between genes can be dramatically improved by incorporating the "HUb effect in Genetic Essentiality" (HUGE) of gene alterations. In three genome-wide loss-of-function screens-Project Score, CERES score and DEMETER score-LEDs are identified with 75 times larger statistical power than using state-of-the-art methods. In AML, we identified LEDs not recalled by previous pipelines, including FLT3-mutant genotypes sensitive to FLT3 inhibitors. Interestingly, in-vitro validations confirm lethal de-pendencies of either NRAS or PTPN11 depending on the NRAS mutational status. Recent functional genomic screens-such as CRISPR-Cas9 or RNAi screening-have fostered a new wave of targeted treatments based on the concept of synthetic lethality. These approaches identified LEthal Dependencies (LEDs) by estimating the effect of genetic events on cell viability. The multiple-hypothesis problem is related to a large number of gene knockouts limiting the statistical power of these studies. Here, we show that predictions of LEDs from functional screens can be dramatically improved by incorporating the "HUb effect in Genetic Essentiality" (HUGE) of gene alterations. We analyze three recent genome-wide loss-of-function screens-Project Score, CERES score and DEMETER score-identifying LEDs with 75 times larger statistical power than using state-of-the-art methods. Using acute myeloid leukemia, breast cancer, lung adenocarcinoma and colon adenocarcinoma as disease models, we validate that our predictions are enriched in a recent harmonized knowledge base of clinical interpretations of somatic genomic variants in cancer (AUROC > 0.87). Our approach is effective even in tumors with large genetic heterogeneity such as acute myeloid leukemia, where we identified LEDs not recalled by previous pipelines, including FLT3-mutant genotypes sensitive to FLT3 inhibitors. Interestingly, in-vitro validations confirm lethal dependencies of either NRAS or PTPN11 depending on the NRAS mutational status. HUGE will hopefully help discover novel genetic dependencies amenable for precision-targeted therapies in cancer. All the graphs showing lethal dependencies for the 19 tumor types analyzed can be visualized in an interactive tool.

Motivation: Isoform deconvolution is an NP-hard problem. The accuracy of the proposed solutions is far from perfect. At present, it is not known if gene structure and isoform concentration can be uniquely inferred given paired-end reads, and there is no objective method to select the fragment length to improve the number of identifiable genes. Different pieces of evidence suggest that the optimal fragment length is gene-dependent, stressing the need for a method that selects the fragment length according to a reasonable trade-off across all the genes in the whole genome. Results: A gene is considered to be identifiable if it is possible to get both the structure and concentration of its transcripts univocally. Here, we present a method to state the identifiability of this deconvolution problem. Assuming a given transcriptome and that the coverage is sufficient to interrogate all junction reads of the transcripts, this method states whether or not a gene is identifiable given the read length and fragment length distribution. Applying this method using different read and fragment length combinations, the optimal average fragment length for the human transcriptome is around 400-600 nt for coding genes and 150-200 nt for long non-coding RNAs. The optimal read length is the largest one that fits in the fragment length. It is also discussed the potential profit of combining several libraries to reconstruct the transcriptome. Combining two libraries of very different fragment lengths results in a significant improvement in gene identifiability.

Motivation: Discover is an algorithm developed to identify mutually exclusive genomic events. Its main contribution is a statistical analysis based on the Poisson-Binomial (PB) distribution to take into account the mutation rate of genes and samples. Discover is very effective for identifying mutually exclusive mutations at the expense of speed in large datasets: the PB is computationally costly to estimate, and checking all the potential mutually exclusive alterations requires millions of tests. Results: We have implemented a new version of the package called Rediscover that implements exact and approximate computations of the PB. Rediscover exact implementation is slightly faster than Discover for large and medium-sized datasets. The approximation is 100-1000 times faster for them making it possible to get results in less than a minute with a standard desktop. The memory footprint is also smaller in Rediscover. The new package is available at CRAN and provides some functions to integrate its usage with other R packages such as maftools and TCGAbiolinks. Availability and implementation: Rediscover is available at CRAN (https://cran.r-project.org/web/packages/ Rediscover/index.html).

Simple Summary Pancreatic cancers are lethal types of cancer. A majority of patients progress to an advanced and metastatic disease, which remains a major clinical problem. Therefore, it is crucial to identify critical regulators to help predict the disease progression and to develop more efficacious therapeutic approaches. In this work we found that an increased expression of the developmental factor SOX9 is associated with metastasis, a poor prognosis and resistance to therapy in pancreatic ductal adenocarcinoma patients and in cell cultures. We also found that this effect is at least in part due to the ability of SOX9 to regulate the activity of stem cell factors, such as BMI1, in addition to those involved in EMT and metastasis. Background: Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers mainly due to spatial obstacles to complete resection, early metastasis and therapy resistance. The molecular events accompanying PDAC progression remain poorly understood. SOX9 is required for maintaining the pancreatic ductal identity and it is involved in the initiation of pancreatic cancer. In addition, SOX9 is a transcription factor linked to stem cell activity and is commonly overexpressed in solid cancers. It cooperates with Snail/Slug to induce epithelial-mesenchymal transition (EMT) during neural development and in diseases such as organ fibrosis or different types of cancer. Methods: We investigated the roles of SOX9 in pancreatic tumor cell plasticity, metastatic dissemination and chemoresistance using pancreatic cancer cell lines as well as mouse embryo fibroblasts. In addition, we characterized the clinical relevance of SOX9 in pancreatic cancer using human biopsies. Results: Gain- and loss-of-function of SOX9 in PDAC cells revealed that high levels of SOX9 increased migration and invasion, and promoted EMT and metastatic dissemination, whilst SOX9 silencing resulted in metastasis inhibition, along with a phenotypic reversion to epithelial features and loss of stemness potential. In both contexts, EMT factors were not altered. Moreover, high levels of SOX9 promoted resistance to gemcitabine. In contrast, overexpression of SOX9 was sufficient to promote metastatic potential in K-Ras transformed MEFs, triggering EMT associated with Snail/Slug activity. In clinical samples, SOX9 expression was analyzed in 198 PDAC cases by immunohistochemistry and in 53 patient derived xenografts (PDXs). SOX9 was overexpressed in primary adenocarcinomas and particularly in metastases. Notably, SOX9 expression correlated with high vimentin and low E-cadherin expression. Conclusions: Our results indicate that SOX9 facilitates PDAC progression and metastasis by triggering stemness and EMT.

Understanding how diet and gut microbiota interact in the context of human health is a key question in personalized nutrition. Genome-scale metabolic networks and constraint-based modeling approaches are promising to systematically address this complex problem. However, when applied to nutritional questions, a major issue in existing reconstructions is the limited information about compounds in the diet that are metabolized by the gut microbiota. Here, we present AGREDA, an extended reconstruction of diet metabolism in the human gut microbiota. AGREDA adds the degradation pathways of 209 compounds present in the human diet, mainly phenolic compounds, a family of metabolites highly relevant for human health and nutrition. We show that AGREDA outperforms existing reconstructions in predicting diet-specific output metabolites from the gut microbiota. Using 16S rRNA gene sequencing data of faecal samples from Spanish children representing different clinical conditions, we illustrate the potential of AGREDA to establish relevant metabolic interactions between diet and gut microbiota. The interplay between human diet and the gut microbiome is complex. Here, the authors present a model of human-microbiome interaction that can predict how phenolic compounds are metabolized by the human gut microbiome, identifying diet-specific metabolites in children of varied clinical conditions.

Direct Yaw Moment Control (DYC) is an effective way to alter the behaviour of electric cars with independent drives. Controlling the torque applied to each wheel can improve the handling performance of a vehicle making it safer and faster on a race track. The state-of-the-art literature covers the comparison of various controllers (PID, LPV, LQR, SMC, etc.) using ISO manoeuvres. However, a more advanced comparison of the important characteristics of the controllers' performance is lacking, such as the robustness of the controllers under changes in the vehicle model, steering behaviour, use of the friction circle, and, ultimately, lap time on a track. In this study, we have compared the controllers according to some of the aforementioned parameters on a modelled race car. Interestingly, best lap times are not provided by perfect neutral or close-to-neutral behaviour of the vehicle, but rather by allowing certain deviations from the target yaw rate. In addition, a modified Proportional Integral Derivative (PID) controller showed that its performance is comparable to other more complex control techniques such as Model Predictive Control (MPC).

The development of predictive biomarkers of response to targeted therapies is an unmet clinical need for many antitumoral agents. Recent genome-wide loss-of-function screens, such as RNA interference (RNAi) and CRISPR-Cas9 libraries, are an unprecedented resource to identify novel drug targets, reposition drugs and associate predictive biomarkers in the context of precision oncology. In this work, we have developed and validated a large-scale bioinformatics tool named DrugSniper, which exploits loss-of-function experiments to model the sensitivity of 6237 inhibitors and predict their corresponding biomarkers of sensitivity in 30 tumor types. Applying DrugSniper to small cell lung cancer (SCLC), we identified genes extensively explored in SCLC, such as Aurora kinases or epigenetic agents. Interestingly, the analysis suggested a remarkable vulnerability to polo-like kinase 1 (PLK1) inhibition inCREBBP-mutant SCLC cells. We validated this association in vitro using four mutated and four wild-type SCLC cell lines and twoPLK1inhibitors (Volasertib and BI2536), confirming that the effect ofPLK1inhibitors depended on the mutational status ofCREBBP. Besides, DrugSniper was validated in-silico with several known clinically-used treatments, including the sensitivity of Tyrosine Kinase Inhibitors (TKIs) and Vemurafenib toFLT3andBRAFmutant cells, respectively. These findings show the potential of genome-wide loss-of-function screens to identify new personalized therapeutic hypotheses in SCLC and potentially in other tumors, which is a valuable starting point for further drug development and drug repositioning projects.

The advent of RNA-seq technologies has switched the paradigm of genetic analysis from a genome to a transcriptome-based perspective. Alternative splicing generates functional diversity in genes, but the precise functions of many individual isoforms are yet to be elucidated. Gene Ontology was developed to annotate gene products according to their biological processes, molecular functions and cellular components. Despite a single gene may have several gene products, most annotations are not isoform-specific and do not distinguish the functions of the different proteins originated from a single gene. Several approaches have tried to automatically annotate ontologies at the isoform level, but this has shown to be a daunting task. We have developed ISOGO (ISOform + GO function imputation), a novel algorithm to predict the function of coding isoforms based on their protein domains and their correlation of expression along 11,373 cancer patients. Combining these two sources of information outperforms previous approaches: it provides an area under precision-recall curve (AUPRC) five times larger than previous attempts and the median AUROC of assigned functions to genes is 0.82. We tested ISOGO predictions on some genes with isoform-specific functions (BRCA1, MADD,VAMP7 and ITSN1) and they were coherent with the literature. Besides, we examined whether the main isoform of each gene -as predicted by APPRIS- was the most likely to have the annotated gene functions and it occurs in 99.4% of the genes. We also evaluated the predictions for isoform-specific functions provided by the CAFA3 challenge and results were also convincing. To make these results available to the scientific community, we have deployed a web application to consult ISOGO predictions (https://biotecnun.unav.es/app/isogo). Initial data, website link, isoform-specific GO function predictions and R code is available at https://gitlab.com/icassol/isogo.

During the last decade, gene expression microarrays and array-based comparative genomic hybridization (array-CGH) have unraveled the complexity of human tumor genomes more precisely and comprehensively than ever before. More recently, the simultaneous assessment of global changes in messenger RNA (mRNA) expression and in DNA copy number through "integrative oncogenomic" analyses has allowed researchers the access to results uncovered through the analysis of one-dimensional data sets, thus accelerating cancer gene discovery. In this chapter, we discuss the major contributions of DNA microarrays to the study of hematological malignancies, focusing on the integrative oncogenomic approaches that correlate genomic and transcriptomic data. We also present the basic aspects of these methodologies and their present and future application in clinical oncology.