CIME4R: Exploring iterative, AI-guided chemical reaction optimization campaigns in their parameter space.

Chemical reaction optimization (RO) is an iterative process that results in large, high-dimensional datasets. Current tools allow for only limited analysis and understanding of parameter spaces, making it hard for scientists to review or follow changes throughout the process. With the recent emergence of using artificial intelligence (AI) models to aid RO, another level of complexity has been added. Helping to assess the quality of a model's prediction and understand its decision is critical to supporting human-AI collaboration and trust calibration. To address this, we propose CIME4R-an open-source interactive web application for analyzing RO data and AI predictions. CIME4R supports users in (i) comprehending a reaction parameter space, (ii) investigating how an RO process developed over iterations, (iii) identifying critical factors of a reaction, and (iv) understanding model predictions. This facilitates making informed decisions during the RO process and helps users to review a completed RO process, especially in AI-guided RO. CIME4R aids decision-making through the interaction between humans and AI by combining the strengths of expert experience and high computational precision. We developed and tested CIME4R with domain experts and verified its usefulness in three case studies. Using CIME4R the experts were able to produce valuable insights from past RO campaigns and to make informed decisions on which experiments to perform next. We believe that CIME4R is the beginning of an open-source community project with the potential to improve the workflow of scientists working in the reaction optimization domain. SCIENTIFIC CONTRIBUTION: To the best of our knowledge, CIME4R is the first open-source interactive web application tailored to the peculiar analysis requirements of reaction optimization (RO) campaigns. Due to the growing use of AI in RO, we developed CIME4R with a special focus on facilitating human-AI collaboration and understanding of AI models. We developed and evaluated CIME4R in collaboration with domain experts to verify its practical usefulness.

XSMILES: interactive visualization for molecules, SMILES and XAI attribution scores.

BACKGROUND: Explainable artificial intelligence (XAI) methods have shown increasing applicability in chemistry. In this context, visualization techniques can highlight regions of a molecule to reveal their influence over a predicted property. For this purpose, some XAI techniques calculate attribution scores associated with tokens of SMILES strings or with atoms of a molecule. While an association of a score with an atom can be directly visually represented on a molecule diagram, scores computed for SMILES non-atom tokens cannot. For instance, a substring [N+] contains 3 non-atom tokens, i.e., [, [Formula: see text], and ], and their attributions, depending on the model, are not necessarily revealing an influence of the nitrogen atom over the predicted property; for that reason, it is not possible to represent the scores on a molecule diagram. Moreover, SMILES's notation is complex, foregrounding the need for techniques to facilitate the analysis of explanations associated with their tokens. RESULTS: We propose XSMILES, an interactive visualization technique, to explore explainable artificial intelligence attributions scores and support the interpretation of SMILES. Users can input any type of score attributed to atom and non-atom tokens and visualize them on top of a 2D molecule diagram coordinated with a bar chart that represents a SMILES string. We demonstrate how attributions calculated for SMILES strings can be evaluated and better interpreted through interactivity with two use cases. CONCLUSIONS: Data scientists can use XSMILES to understand their models' behavior and compare multiple modeling approaches. The tool provides a set of parameters to adapt the visualization to users' needs and it can be integrated into different platforms. We believe XSMILES can support data scientists to develop, improve, and communicate their models by making it easier to identify patterns and compare attributions through interactive exploratory visualization.

Connecting multiple microenvironment proteomes uncovers the biology in head and neck cancer.

The poor prognosis of head and neck cancer (HNC) is associated with metastasis within the lymph nodes (LNs). Herein, the proteome of 140 multisite samples from a 59-HNC patient cohort, including primary and matched LN-negative or -positive tissues, saliva, and blood cells, reveals insights into the biology and potential metastasis biomarkers that may assist in clinical decision-making. Protein profiles are strictly associated with immune modulation across datasets, and this provides the basis for investigating immune markers associated with metastasis. The proteome of LN metastatic cells recapitulates the proteome of the primary tumor sites. Conversely, the LN microenvironment proteome highlights the candidate prognostic markers. By integrating prioritized peptide, protein, and transcript levels with machine learning models, we identify nodal metastasis signatures in blood and saliva. We present a proteomic characterization wiring multiple sites in HNC, thus providing a promising basis for understanding tumoral biology and identifying metastasis-associated signatures.

ChemInformatics Model Explorer (CIME): exploratory analysis of chemical model explanations.

The introduction of machine learning to small molecule research- an inherently multidisciplinary field in which chemists and data scientists combine their expertise and collaborate - has been vital to making screening processes more efficient. In recent years, numerous models that predict pharmacokinetic properties or bioactivity have been published, and these are used on a daily basis by chemists to make decisions and prioritize ideas. The emerging field of explainable artificial intelligence is opening up new possibilities for understanding the reasoning that underlies a model. In small molecule research, this means relating contributions of substructures of compounds to their predicted properties, which in turn also allows the areas of the compounds that have the greatest influence on the outcome to be identified. However, there is no interactive visualization tool that facilitates such interdisciplinary collaborations towards interpretability of machine learning models for small molecules. To fill this gap, we present CIME (ChemInformatics Model Explorer), an interactive web-based system that allows users to inspect chemical data sets, visualize model explanations, compare interpretability techniques, and explore subgroups of compounds. The tool is model-agnostic and can be run on a server or a workstation.

Combining discovery and targeted proteomics reveals a prognostic signature in oral cancer.

Different regions of oral squamous cell carcinoma (OSCC) have particular histopathological and molecular characteristics limiting the standard tumor-node-metastasis prognosis classification. Therefore, defining biological signatures that allow assessing the prognostic outcomes for OSCC patients would be of great clinical significance. Using histopathology-guided discovery proteomics, we analyze neoplastic islands and stroma from the invasive tumor front (ITF) and inner tumor to identify differentially expressed proteins. Potential signature proteins are prioritized and further investigated by immunohistochemistry (IHC) and targeted proteomics. IHC indicates low expression of cystatin-B in neoplastic islands from the ITF as an independent marker for local recurrence. Targeted proteomics analysis of the prioritized proteins in saliva, combined with machine-learning methods, highlights a peptide-based signature as the most powerful predictor to distinguish patients with and without lymph node metastasis. In summary, we identify a robust signature, which may enhance prognostic decisions in OSCC and better guide treatment to reduce tumor recurrence or lymph node metastasis.

CellNetVis: a web tool for visualization of biological networks using force-directed layout constrained by cellular components.

BACKGROUND: The advent of "omics" science has brought new perspectives in contemporary biology through the high-throughput analyses of molecular interactions, providing new clues in protein/gene function and in the organization of biological pathways. Biomolecular interaction networks, or graphs, are simple abstract representations where the components of a cell (e.g. proteins, metabolites etc.) are represented by nodes and their interactions are represented by edges. An appropriate visualization of data is crucial for understanding such networks, since pathways are related to functions that occur in specific regions of the cell. The force-directed layout is an important and widely used technique to draw networks according to their topologies. Placing the networks into cellular compartments helps to quickly identify where network elements are located and, more specifically, concentrated. Currently, only a few tools provide the capability of visually organizing networks by cellular compartments. Most of them cannot handle large and dense networks. Even for small networks with hundreds of nodes the available tools are not able to reposition the network while the user is interacting, limiting the visual exploration capability. RESULTS: Here we propose CellNetVis, a web tool to easily display biological networks in a cell diagram employing a constrained force-directed layout algorithm. The tool is freely available and open-source. It was originally designed for networks generated by the Integrated Interactome System and can be used with networks from others databases, like InnateDB. CONCLUSIONS: CellNetVis has demonstrated to be applicable for dynamic investigation of complex networks over a consistent representation of a cell on the Web, with capabilities not matched elsewhere.

Integrative analysis to select cancer candidate biomarkers to targeted validation.

Targeted proteomics has flourished as the method of choice for prospecting for and validating potential candidate biomarkers in many diseases. However, challenges still remain due to the lack of standardized routines that can prioritize a limited number of proteins to be further validated in human samples. To help researchers identify candidate biomarkers that best characterize their samples under study, a well-designed integrative analysis pipeline, comprising MS-based discovery, feature selection methods, clustering techniques, bioinformatic analyses and targeted approaches was performed using discovery-based proteomic data from the secretomes of three classes of human cell lines (carcinoma, melanoma and non-cancerous). Three feature selection algorithms, namely, Beta-binomial, Nearest Shrunken Centroids (NSC), and Support Vector Machine-Recursive Features Elimination (SVM-RFE), indicated a panel of 137 candidate biomarkers for carcinoma and 271 for melanoma, which were differentially abundant between the tumor classes. We further tested the strength of the pipeline in selecting candidate biomarkers by immunoblotting, human tissue microarrays, label-free targeted MS and functional experiments. In conclusion, the proposed integrative analysis was able to pre-qualify and prioritize candidate biomarkers from discovery-based proteomics to targeted MS.

InteractiVenn: a web-based tool for the analysis of sets through Venn diagrams.

BACKGROUND: Set comparisons permeate a large number of data analysis workflows, in particular workflows in biological sciences. Venn diagrams are frequently employed for such analysis but current tools are limited. RESULTS: We have developed InteractiVenn, a more flexible tool for interacting with Venn diagrams including up to six sets. It offers a clean interface for Venn diagram construction and enables analysis of set unions while preserving the shape of the diagram. Set unions are useful to reveal differences and similarities among sets and may be guided in our tool by a tree or by a list of set unions. The tool also allows obtaining subsets' elements, saving and loading sets for further analyses, and exporting the diagram in vector and image formats. InteractiVenn has been used to analyze two biological datasets, but it may serve set analysis in a broad range of domains. CONCLUSIONS: InteractiVenn allows set unions in Venn diagrams to be explored thoroughly, by consequence extending the ability to analyze combinations of sets with additional observations, yielded by novel interactions between joined sets. InteractiVenn is freely available online at: www.interactivenn.net .