A benchmark for evaluation of algorithms for identification of cellular correlates of clinical outcomes.

The Flow Cytometry: Critical Assessment of Population Identification Methods (FlowCAP) challenges were established to compare the performance of computational methods for identifying cell populations in multidimensional flow cytometry data. Here we report the results of FlowCAP-IV where algorithms from seven different research groups predicted the time to progression to AIDS among a cohort of 384 HIV+ subjects, using antigen-stimulated peripheral blood mononuclear cell (PBMC) samples analyzed with a 14-color staining panel. Two approaches (FlowReMi.1 and flowDensity-flowType-RchyOptimyx) provided statistically significant predictive value in the blinded test set. Manual validation of submitted results indicated that unbiased analysis of single cell phenotypes could reveal unexpected cell types that correlated with outcomes of interest in high dimensional flow cytometry datasets.

Reduced gene templates for supervised analysis of scale-limited CRISPR-Cas9 fitness screens.

Pooled genome-wide CRISPR-Cas9 screens are furthering our mechanistic understanding of human biology and have allowed us to identify new oncology therapeutic targets. Scale-limited CRISPR-Cas9 screens-typically employing guide RNA libraries targeting subsets of functionally related genes, biological pathways, or portions of the druggable genome-constitute an optimal setting for investigating narrow hypotheses and are easier to execute on complex models, such as organoids and in vivo models. Different supervised methods are used for computational analysis of genome-wide CRISPR-Cas9 screens; most are not well suited for scale-limited screens, as they require large sets of positive/negative control genes (gene templates) to be included among the screened ones. Here, we develop a computational framework identifying optimal subsets of known essential and nonessential genes (at different subsampling percentages) that can be used as templates for supervised analyses of scale-limited CRISPR-Cas9 screens, while having a reduced impact on the size of the employed library.

Integration of Proteomics and Other Omics Data.

In recent biomedical studies, multidimensional profiling, which collects proteomics as well as other types of omics data on the same subjects, is getting increasingly popular. Proteomics, transcriptomics, genomics, epigenomics, and other types of data contain overlapping as well as independent information, which suggests the possibility of integrating multiple types of data to generate more reliable findings/models with better classification/prediction performance. In this chapter, a selective review is conducted on recent data integration techniques for both unsupervised and supervised analysis. The main objective is to provide the "big picture" of data integration that involves proteomics data and discuss the "intuition" beneath the recently developed approaches without invoking too many mathematical details. Potential pitfalls and possible directions for future developments are also discussed.

Elaboration Pipeline for the Management of MALDI-MS Imaging Datasets.

Matrix-assisted laser desorption/ionization (MALDI)-time of flight (TOF)-mass spectrometry imaging (MSI) enables the spatial localization of proteins to be mapped directly on tissue sections, simultaneously detecting hundreds in a single analysis. However, the large data size, as well as the complexity of MALDI-MSI proteomics datasets, requires the appropriate tools and statistical approaches in order to reduce the complexity and mine the dataset in a successful manner. Here, a pipeline for the management of MALDI-MSI data is described, starting with preprocessing of the raw data, followed by statistical analysis using both supervised and unsupervised statistical approaches and, finally, annotation of those discriminatory protein signals highlighted by the data mining procedure.

Supervised Analysis for Phenotype Identification: The Case of Heart Failure Ejection Fraction Class.

Artificial Intelligence is creating a paradigm shift in health care, with phenotyping patients through clustering techniques being one of the areas of interest. OBJECTIVE: To develop a predictive model to classify heart failure (HF) patients according to their left ventricular ejection fraction (LVEF), by using available data from Electronic Health Records (EHR). SUBJECTS AND METHODS: 2854 subjects over 25 years old with a diagnosis of HF and LVEF, measured by echocardiography, were selected to develop an algorithm to predict patients with reduced EF using supervised analysis. The performance of the developed algorithm was tested in heart failure patients from Primary Care. To select the most influentual variables, the LASSO algorithm setting was used, and to tackle the issue of one class exceeding the other one by a large amount, we used the Synthetic Minority Oversampling Technique (SMOTE). Finally, Random Forest (RF) and XGBoost models were constructed. RESULTS: The full XGBoost model obtained the maximum accuracy, a high negative predictive value, and the highest positive predictive value. Gender, age, unstable angina, atrial fibrillation and acute myocardial infarct are the variables that most influence EF value. Applied in the EHR dataset, with a total of 25,594 patients with an ICD-code of HF and no regular follow-up in cardiology clinics, 6170 (21.1%) were identified as pertaining to the reduced EF group. CONCLUSION: The obtained algorithm was able to identify a number of HF patients with reduced ejection fraction, who could benefit from a protocol with a strong possibility of success. Furthermore, the methodology can be used for studies using data extracted from the Electronic Health Records.

Investigation of the limit of discovery using tile-based Fisher ratio analysis with comprehensive two-dimensional gas chromatography time-of-flight mass spectrometry.

Comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (GC×GC-TOFMS) is followed by tile-based Fisher ratio (F-ratio) analysis to investigate the "limit of discovery" for low concentration levels of sulfur-containing compounds in JP8 jet fuel. A mixture of 14 sulfur-containing compounds was spiked at 30 ppm, 15 ppm, 3 ppm and 1.5 ppm into the neat fuel prior to GC×GC-TOFMS analysis with a "reversed" column format (aka polar first dimension (1D) and non-polar second dimension (2D) column). Prior standard implementation of tile-based F-ratio analysis utilized an average F-ratio requiring a minimum of 3 mass channels (m/z) with the highest F-ratios. Herein, we explore the notion that use of the top F-ratio m/z for hitlist ranking is superior to the standard implementation for analytes near their limit-of-quantitation (LOQ), defined as an analyte concentration that produces a signal equal to ten times the standard deviation of the baseline noise (10σn). Hitlist ranking comparisons revealed that using only the top F-ratio m/z resulted in impressive improvements in discoverability for the low concentration comparisons. Specifically, for the 3 ppm versus neat hitlist, 1,4-oxathiane (LOQ = 2.5 ppm) improved from hit 114 via standard F-ratio analysis, to hit 25. For the 1.5 ppm versus neat hitlist, 2-propylthiophene (LOQ = 0.64 ppm) improved from hit 59 to 17, benzo[b]thiophene (LOQ = 1.1 ppm) from hit 98 to 28, and 2,5-dimethylthiophene (LOQ = 1.3 ppm) from hit 262 to 39. Additional hitlist ranking comparisons revealed the importance of proper tile size selection, as analyte discoverability deteriorated upon using either an inappropriately too small or too large of a tile.

Robust model selection for classification of microarrays.

Recently, microarray-based cancer diagnosis systems have been increasingly investigated. However, cost reduction and reliability assurance of such diagnosis systems are still remaining problems in real clinical scenes. To reduce the cost, we need a supervised classifier involving the smallest number of genes, as long as the classifier is sufficiently reliable. To achieve a reliable classifier, we should assess candidate classifiers and select the best one. In the selection process of the best classifier, however, the assessment criterion must involve large variance because of limited number of samples and non-negligible observation noise. Therefore, even if a classifier with a very small number of genes exhibited the smallest leave-one-out cross-validation (LOO) error rate, it would not necessarily be reliable because classifiers based on a small number of genes tend to show large variance. We propose a robust model selection criterion, the min-max criterion, based on a resampling bootstrap simulation to assess the variance of estimation of classification error rates. We applied our assessment framework to four published real gene expression datasets and one synthetic dataset. We found that a state-of-the-art procedure, weighted voting classifiers with LOO criterion, had a non-negligible risk of selecting extremely poor classifiers and, on the other hand, that the new min-max criterion could eliminate that risk. These finding suggests that our criterion presents a safer procedure to design a practical cancer diagnosis system.