Machine Learning Model Analysis and Data Visualization with Small Molecules Tested in a Mouse Model of Mycobacterium tuberculosis Infection (2014–2015)

Abstract

The renewed urgency to develop new treatments for Mycobacterium tuberculosis (Mtb) infection has resulted in large-scale phenotypic screening and thousands of new active compounds in vitro. The next challenge is to identify candidates to pursue in a mouse in vivo efficacy model as a step to predicting clinical efficacy. We previously analyzed over 70 years of this mouse in vivo efficacy data, which we used to generate and validate machine learning models. Curation of 60 additional small molecules with in vivo data published in 2014 and 2015 was undertaken to further test these models. This represents a much larger test set than for the previous models. Several computational approaches have now been applied to analyze these molecules and compare their molecular properties beyond those attempted previously. Our previous machine learning models have been updated, and a novel aspect has been added in the form of mouse liver microsomal half-life (MLM t1/2) and in vitro-based Mtb models incorporating cytotoxicity data that were used to predict in vivo activity for comparison. Our best Mtbin vivo models possess fivefold ROC values > 0.7, sensitivity > 80%, and concordance > 60%, while the best specificity value is >40%. Use of an MLM t1/2 Bayesian model affords comparable results for scoring the 60 compounds tested. Combining MLM stability and in vitroMtb models in a novel consensus workflow in the best cases has a positive predicted value (hit rate) > 77%. Our results indicate that Bayesian models constructed with literature in vivoMtb data generated by different laboratories in various mouse models can have predictive value and may be used alongside MLM t1/2 and in vitro-based Mtb models to assist in selecting antitubercular compounds with desirable in vivo efficacy. We demonstrate for the first time that consensus models of any kind can be used to predict in vivo activity for Mtb. In addition, we describe a new clustering method for data visualization and apply this to the in vivo training and test data, ultimately making the method accessible in a mobile app.

Learning from the past for TB drug discovery in the future

“…Use the computational models to identify compounds that could be active in vivo [67,69] from the 1000s of published in vitro hits.…”

“…The creation of this consensus workflow had a positive predicted value (hit rate) >77%. Such computational models can help select antitubercular compounds with desirable in vivo efficacy alongside good MLM t 1/2 and in vitro activity [69]. We also found that a new clustering method for data visualization might also help with the assessment of in vivo activity by placing a new molecule of interest in the context of the closest molecules based on molecular descriptors (Figure 1).…”

“…Certainly the mouse differs from the human but it represents a frequently used model for rank ordering compounds for further study [68]. Recent efforts have manually collated historic data for compounds tested in the mouse model of infection [67,69]. Admittedly this represents a combination of data from different strains of mice and acute and chronic mouse models covering over 70 years of data.…”

“…Our first analysis also enabled us to take a high-level view of the compounds assessed in the mouse over time [4] and the clear gaps in drug discovery. The more-recent curation of 60 additional small molecules with in vivo data published during 2014 and 2015 [69] suggests a general lack of diversity of compounds going in vivo . We have also used multiple computational approaches to analyze these molecules and we showed that models to predict other properties such as mouse liver microsomal half-life (MLM t 1/2 ) and in vitro M. tuberculosis activity incorporating cytotoxicity data could also be used to predict in vivo activity.…”

“…We identified this because we had curated over 700 molecules with their potential targets from literature publications in a mobile app called TB Mobile [77,78]. Resources like this [79,80], mouse in vivo compounds [67,69] and literature in vitro screening data from SRI/NIAID [81-83] that has been collated in CDD (http://www.collaborativedrug.com) [11,84] all represent starting points for potential knowledge-based selection of compounds for further follow-up or design. Using the computational models or the data on their own can help in understanding the properties that confer in vitro activity , in vivo activity as well as the multidimensional properties such as cytotoxicity and microsomal stability (Figure 3).…”