XTalkiiS: a tool for finding data-driven cross-talks between intra-/inter-species pathways

Cell-cell communication via pathway cross-talks within a single species have been studied in silico recently to decipher various disease phenotype. However, computational prediction of pathway cross-talks among multiple species in a data-driven manner is yet to be explored. In this article, I present XTalkiiS (Cross-talks between inter-/intra species pathways), a tool to automatically predict pathway cross-talks from data-driven models of pathway network, both within the same organism (intra-species) and between two organisms (inter-species). XTalkiiS starts with retrieving and listing up-to-date pathway information in all the species available in KEGG database using RESTful APIs (exploiting KEGG web services) and an in-house built web crawler. I hypothesize that data-driven network models can be built by simultaneously quantifying co-expression of pathway components (i.e. genes/proteins) in matched samples in multiple organisms. Next, XTalkiiS loads a data-driven pathway network and applies a novel cross-talk modelling approach to determine interactions among known KEGG pathways in selected organisms. The potentials of XTalkiiS are huge as it paves the way of finding novel insights into mechanisms how pathways from two species (ideally host-parasite) may interact that may contribute to the various phenotype of interests such as malaria disease. XTalkiiS is made open sourced at https://github.com/Akmazad/XTalkiiS and its binary files are freely available for downloading from https://sourceforge.net/projects/xtalkiis/.

In biological systems, genes doesn't work alone; rather as a group they perform certain 2 activities named as pathways. Sometimes, the activities of certain pathways even gets 3 modified by others pathways due to the interference among them, a phenomenon called 4 'pathway cross-talks'. In many cases such interference plays critical roles in mediating 5 novel mechanisms of certain diseases-associated activities e.g. tumor progression or 6 acquired drug resistance. Moreover, these cross-talks may also be present not only 7 within the same organism but also between organisms e.g. interference between 8 host-parasite pathway activities. 9 Development of many complex diseases such as cancer often happens due to the 10 genetic/epigenetic alteration of some key driver genes and their perturbed influence 11 through their pathway activities [5]. Moreover, recent researches in vivo and in vitro 12 have shown that cancer cells acquires resistance to particular inhibitors by adapting 13 1/5 their signalling circuitry, activation of alternate pathways and cross-talks among various 14 signalling pathways [5]. Again, Komurov et al. reported that regulatory pathways such 15 as glucose deprivation response pathway cross-talks with EGFR-mediated pathways to 16 provide cancer cells an alternate route for glucose intake, which is essential for their 17 survival for which tumor ultimately relapses [9]. So, for better understanding of 18 pathway activities within a biological systems under various research hypotheses, it is 19 often crucial to study the pathway cross-talks within the organism.

20
Moreover, it is important to study interactions among multiple organisms, especially 21 within microbial ecosystems [10] for several reasons: 1) how their pathways are 22 interconnected, 2) how their interactions changes due the perturbation of one or both of 23 their systems, 3) how do they correspond to external factors such environmental 24 changes or treatment stimuli, etc. Hence, more research required to determine and 25 characterize these inter-species pathway cross-talks to reveal better insights into the 26 networks of diseases-associated biological pathways in a data-driven manner.

27
Gene or protein co-expression measurements using high-throughput data sets can 28 model their relationships/dependencies in performing a particular biological activities [6] 29 in a particular pathway in a data-driven manner. These putative relationships can form 30 a network structure mimicking pathways [6], and their interactions in a data-driven 31 manner [5], where the nodes are individual genes/proteins and edges are the 32 relationships among them within that particular species. I hypothesize that similar 33 approach could be applied to study the relationships among pathway components from 34 multiple species by forming a mega network structure, which can be determined by, 1) 35 first measuring genome-wide high-throughput information of pathway genes/proteins 36 simultaneously in multiple organism, and then 2) evaluating their co-expression.

37
Data-driven models of pathway network provide useful ways of capturing dynamic 38 patterns in pathway activities in a context-specific manner [2]. However, typical 39 definitions of pathway cross-talks aren't suitable for data-driven models of pathway 40 networks as they may have novel dependencies among genes/proteins within those 41 network structure [2]. Hence, we defined a novel cross-talk categorization namely Type-I 42 and Type-II cross-talks that are suitable for data-driven network models and covers all 43 the cross-talk definitions from the state-of-the-art approaches [2].

44
In this study, we proposed a framework and a software tool called XTalkiiS 45 (Cross-talks between inter-/intra species pathways) that uses our previously proposed 46 cross-talk modeling suitable for data-driven pathway networks [2] and determine 47 pathway cross-talks within the same species (intra-species) and between two-component 48 biological systems i.e. host-parasite manner. 49 Implementation 50 Figure 1 demonstrates the main interface of XTalkiiS. XTalkiiS operates to find pathway 51 cross-talks within the same species (or organism) or between two species. Upon loading 52 the main interface, XTalkiiS retrieves all the organism names from KEGG database by 53 making HTTP web request via a RESTful API call, and lists them in two combo-boxes 54 side-by-side from where users can select two organisms. Upon selecting an organism 55 from each of these two combo-boxes, all pathways names available in the KEGG 56 databases for corresponding organisms are again retrieved using another RESTful API 57 call. When all the pathways are listed for those two organisms, some (or all) of them can 58 be selected which populate two corresponding list-boxes. Next, user loads a gene-gene 59 dependency file from local machine which has information of following structure, Then upon pressing a button it will retrieve all the up-to-date gene names in all the 67 pathways in those two list-boxes via another RESTful API call, which is similar to a 68 previously developed tool called KPGminer [4]. These API calling protocols are listed in 69 the KEGG website for developers' use. In the same event, it applies the cross-talk 70 categorizations [2] on the gene-gene dependency file loaded before and finds two types of 71 cross-talks, Type-I and Type-II cross-talks. Next, the output panel on the right shows, 72 1) all the pathway genes in the first grid-view, and 2) all the Type-I and Type-II 73 cross-talks among all the pathways in intra-/inter-species manner in the second 74 grid-view. Next, I apply XTalkiiS on gene-gene dependency data set collected from [3], which is a 77 list of aberrant gene-pairs derived from lapatinib-sensitive and lapatinib-resistant BT474 78 3/5 cell-lines [GSE16179] using Bayesian statistical modelling [for details see Methods 79 section from [3]]. This example demonstrates the data-driven pathway cross-talks from 80 human breast cancer samples, hence indicating intra-species pathway cross-talks. This  This version of XTalkiiS has one limitation: it loads pathway genes for each 96 individual pathways via making HTTP web request iteratively for each pathways which 97 may be time-consuming depending on the number of pathways. In the next versions, it 98 will adapt multi-threading approach where each thread will make separate HTTP web 99 request, especially because the gene loading step for each pathways are independent of 100 each others. Moreover, XTalkiiS can be resourced with more pathway databases 101 including Reactome [7], Wikipathways [8] or GO terms [1]. I hope, XTalkiiS will be a 102 resourceful tool in future, when finding inter-species pathway cross-talks will become a 103 necessary research component in the field of computational biology.