SUMMARY: Brain Predictability toolbox (BPt) represents a unified framework of machine learning (ML) tools designed to work with both tabulated data (e.g. brain derived, psychiatric, behavioral and physiological variables) and neuroimaging specific data (e.g. brain volumes and surfaces). This package is suitable for investigating a wide range of different neuroimaging-based ML questions, in particular, those queried from large human datasets. AVAILABILITY AND IMPLEMENTATION: BPt has been developed as an open-source Python 3.6+ package hosted at https://github.com/sahahn/BPt under MIT License, with documentation provided at https://bpt.readthedocs.io/en/latest/, and continues to be actively developed. The project can be downloaded through the github link provided. A web GUI interface based on the same code is currently under development and can be set up through docker with instructions at https://github.com/sahahn/BPt_app.

MOTIVATION: Network diffusion and label propagation are fundamental tools in computational biology, with applications like gene-disease association, protein function prediction and module discovery. More recently, several publications have introduced a permutation analysis after the propagation process, due to concerns that network topology can bias diffusion scores. This opens the question of the statistical properties and the presence of bias of such diffusion processes in each of its applications. In this work, we characterized some common null models behind the permutation analysis and the statistical properties of the diffusion scores. We benchmarked seven diffusion scores on three case studies: synthetic signals on a yeast interactome, simulated differential gene expression on a protein-protein interaction network and prospective gene set prediction on another interaction network. For clarity, all the datasets were based on binary labels, but we also present theoretical results for quantitative labels. RESULTS: Diffusion scores starting from binary labels were affected by the label codification and exhibited a problem-dependent topological bias that could be removed by the statistical normalization. Parametric and non-parametric normalization addressed both points by being codification-independent and by equalizing the bias. We identified and quantified two sources of bias-mean value and variance-that yielded performance differences when normalizing the scores. We provided closed formulae for both and showed how the null covariance is related to the spectral properties of the graph. Despite none of the proposed scores systematically outperformed the others, normalization was preferred when the sought positive labels were not aligned with the bias. We conclude that the decision on bias removal should be problem and data-driven, i.e. based on a quantitative analysis of the bias and its relation to the positive entities. AVAILABILITY: The code is publicly available at https://github.com/b2slab/diffuBench and the data underlying this article are available at https://github.com/b2slab/retroData. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Summary: Label propagation and diffusion over biological networks are a common mathematical formalism in computational biology for giving context to molecular entities and prioritizing novel candidates in the area of study. There are several choices in conceiving the diffusion process-involving the graph kernel, the score definitions and the presence of a posterior statistical normalization-which have an impact on the results. This manuscript describes diffuStats, an R package that provides a collection of graph kernels and diffusion scores, as well as a parallel permutation analysis for the normalized scores, that eases the computation of the scores and their benchmarking for an optimal choice. Availability and implementation: The R package diffuStats is publicly available in Bioconductor, https://bioconductor.org, under the GPL-3 license. Contact: [email protected]. Supplementary information: Supplementary data are available at Bioinformatics online.

MOTIVATION: Genome-wide association studies (GWAS) have largely failed to identify most of the genetic basis of highly heritable diseases and complex traits. Recent work has suggested this could be because many genetic variants, each with individually small effects, compose their genetic architecture, limiting the power of GWAS, given currently obtainable sample sizes. In this scenario, Bonferroni-derived thresholds are severely underpowered to detect the vast majority of associations. Local false discovery rate (fdr) methods provide more power to detect non-null associations, but implicit assumptions about the exchangeability of single nucleotide polymorphisms (SNPs) limit their ability to discover non-null loci. METHODS: We propose a novel covariate-modulated local false discovery rate (cmfdr) that incorporates prior information about gene element-based functional annotations of SNPs, so that SNPs from categories enriched for non-null associations have a lower fdr for a given value of a test statistic than SNPs in unenriched categories. This readjustment of fdr based on functional annotations is achieved empirically by fitting a covariate-modulated parametric two-group mixture model. The proposed cmfdr methodology is applied to a large Crohn's disease GWAS. RESULTS: Use of cmfdr dramatically improves power, e.g. increasing the number of loci declared significant at the 0.05 fdr level by a factor of 5.4. We also demonstrate that SNPs were declared significant using cmfdr compared with usual fdr replicate in much higher numbers, while maintaining similar replication rates for a given fdr cutoff in de novo samples, using the eight Crohn's disease substudies as independent training and test datasets. Availability an implementation: https://sites.google.com/site/covmodfdr/ CONTACT: : [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.