TuxNet: A simple interface to process RNA sequencing data and infer gene regulatory networks

Abstract

14 performed the experimental work; R.S. designed and supervised the experiments, and complemented the writing. 15 16 SUMMARY 17 TuxNet offers a simple interface for non-computational biologists to infer GRNs from raw RNA-18 seq data.19 20 ABSTRACT 21 22Predicting gene regulatory networks (GRNs) from gene expression profiles has become a 23 common approach for identifying important biological regulators. Despite the increase in the use 24 of inference methods, existing computational approaches do not integrate RNA-sequencing 25 data analysis, are often not automated, and are restricted to users with bioinformatics and 26 programming backgrounds. To address these limitations, we have developed TuxNet, an 27 integrated user-friendly platform, which, with just a few selections, allows to process raw RNA-28 sequencing data (using the Tuxedo pipeline) and infer GRNs from these processed data. 29TuxNet is implemented as a graphical user interface and, using expression data from any 30 organism with an existing reference genome, can mine the regulations among genes either by 31 applying a dynamic Bayesian network inference algorithm, GENIST, or a regression tree-based 32 pipeline that uses spatiotemporal data, RTP-STAR. To illustrate the use of TuxNet while getting 33 insight into the regulatory cascade downstream of the Arabidopsis root stem cell regulator 34 PERIANTHIA (PAN), we obtained time course gene expression data of a PAN inducible line and 35 inferred a GRN using GENIST. Using RTP-STAR, we then inferred the network of a PAN 36 secondary downstream gene, ATHB13, for which we obtained wildtype and mutant expression 37 profiles. Our case studies feature the versatility of TuxNet to infer networks using different types 38 of gene expression data (i.e time course and steady-state data) as well as how inference 39 networks are used to identify important regulators. 40 41 42 81 gene, ATHB13, were used to infer a GRN using RTP-STAR. Importantly, through our examples, 82 we guide the user on how to select the few parameters needed to run the GUI, which we 83 provide as an open source package (https://github.com/madeluis/TuxNet). 84 85 86 87 3 RESULTS 89Overview of the three tabs of TuxNet 90TuxNet provides a user-friendly environment where users can automatically process gene 91 expression data (RNA-seq data), identify differentially expressed genes (DEGs), and infer gene 92 regulatory networks (GRNs) from the processed data. TuxNet is developed as a graphical user 93 interface (GUI) divided in three tabs, TUX, GENIST, and RTP-STAR ( Fig. 1, Fig. 2, and Fig. 3). 94The three TuxNet tabs have consistent file formatting to ensure that analyses, when expanding 95 more than one tab, do not require data manipulation or coding by the user. 97The TUX tab allows users to process RNA-seq data in the form of fastq files, as usually returned 98 by an Illumina sequencer. Accordingly, the TUX tab, which implements fastq-mcf (Aronesty 99 2011; Aronesty 2013) and the Tuxedo pipeline (TopHat + Cufflinks + Cuffdiff) (Trapnell et al. 100...

Stem-cell-ubiquitous genes spatiotemporally coordinate division through regulation of stem-cell-specific gene networks

“…To test the latter hypothesis, in which stem-cell-specific genes are important for regulating cell type-specific aspects (e.g cell identity), but are regulated by stem-cell-ubiquitous genes so that stem cell maintenance and divisions are tightly coordinated, we used Gene Regulatory Network (GRN) inference and predicted the relationships between all 9266 genes enriched in the stem cells. We used a machine-learning, regression tree approach to infer dynamic networks from steady state, replicate data 11 . Our inferred GRN found regulations among 2982 (32.2%) of the stem cell-enriched genes and predicted that the stem-cell-ubiquitous (red) genes are located in the center of the network, which represents the beginning of the regulatory cascade, and are highly connected to each other (Figure 2A).…”

“…The Regression Tree Pipeline for Spatial, Temporal, and Replicate data (RTP-STAR 11 ) was used for all network inference. All networks were inferred using the default parameters described in 11 .…”

“…The Regression Tree Pipeline for Spatial, Temporal, and Replicate data (RTP-STAR 11 ) was used for all network inference. All networks were inferred using the default parameters described in 11 . For the SCN GRN, networks were inferred for each stem cell separately (resulting in 6 networks, one for each stem cell) and then combined to form the final network.…”

Positive feedback and feedforward loops between PERIANTHIA, WUSCHEL-RELATED HOMEOBOX5 and GRF-INTERACTING FACTOR 1 modulate gene expression and function in the Arabidopsis root

“…To determine if WOX5 regulates genes similarly in the QC and CEI, we focused on the 200 genes (16 TFs) identifies as DE in both the QC and CEI of wox5-1 roots compared to WT (Supplemental Table 5 , 6). Moreover, to better understand the relationship between the 16 DE TFs, we inferred a Gene Regulatory Network (GRN) using RTP-STAR as inference algorithm (de Luis Balaguer, 2018;Shibata et al, 2018). The application of RTP-STAR with our QC and CEI cell-type resulted in an inferred network of 12 TFs (Figure 2A), for which we performed a network motif analysis to rank the functionally important factors.…”