USER MANUAL - National Academy of Sciences of Armenia
Contents
1. PostProcessed graph Graphi nodes 0 Node ID 0 name PKCdelta value 1 0 level 0 signals 1 0 1 0 1 0 1 0 1 0 1 0 1 0 The signals of the apoptosis 1 Node ID 1 name IGFiR value 0 1 level 0 signals 0 1 0 1 0 1 0 1 0 1 0 1 0 1 2 Node ID 2 name p S6 value 1 0 level 3 signals 1 0 0 97 0 9 0 95 0 95 0 9 z 3 Node ID 3 name p P70S6K value 1 0 level 2 signals 1 0 0 97 24 r O 1 07 1 12 1 1 1 11 1 1 1 1 Jy e e e O node after each iteration 4 Node ID 4 name apoptosis value 1 0 level i1 signals S Node ID 5 name mTOR value 1 0 level 0 signals 1 0 6 Node ID 6 name Gi arrest value 1 0 level 2 signals 1 29 1 36 1 41 1 4 1 4 1 39 1 39 7 Node ID 7 name p ERK value 1 0 level 3 signals 0 87 0 97 0 95 0 98 0 96 0 97 0 96 8 Node ID 8 name p AKT value 1 0 level 4 signals 0 87 0 79 0 81 0 81 0 82 0 82 0 82 9 Node ID 9 name PI3K value 1 0 level 3 signals 0 87 0 79 0 81 0 81 0 82 0 82 0 82 10 Node ID 10 name p elIF4E value 1 0 level 4 signals 0 87 0 97 0 95 0 98 0 96 0 97 0 96 11 Node ID 11 name p RAF value 1 0 level 6 signals 1 02 1 01 1 02 1 02 1 02 1 02 1 02 12 Node ID 12 name EGFR value 1 0 level 4 signals 1 23 1 22 1 27 1 24 1 25 1 2 1 25 i13 Node ID 13 name p MEK value 1 0 level 5 signals 1 23 1 22 1 27 1 2 1 25 1 2 1 25 edges Edge source ID 0 name PKCdelta target2. SA Y a i Je i g after sorting EN Network layout changed Input nodes have level 0 PSF calculation proceeds from lower level nodes to higher ones Currently only a modified version topological sort algorithm is provided as an option For directed acyclic graphs DAGs topological sorting algorithm orders the vertices of a graph in a way that for every edge uv for nodes u and v u comes before v in the ordering The algorithm uses J GraphT s TopologicalOrderlterator but considers the fact that the graph may have more than one input vertex Biological pathways however mostly contain cycles that represent positive or negative feedback loops For sorting cycle containing graphs we first remove backward edges which are found with a modified depth first search algorithm which also accounts for multiple input nodes After removing the backward edges the resulting graph is sorted with topological sorting algorithm and finally the backward edges are restored in the graph 24 Flow propagation rules In biological signaling networks functional interaction types can be broadly defined as activation or inhibition while the range of physical and regulatory interactions is much wider phosphorylation binding dimerization ubiquitination etc Because of different types of interactions each edge in a pathway should be associated with a different function that will describe how signal will propaga
3. vs proportional m gt 20 edges Multiple signal processing Multiple signals entering a single target node may be processed in one the following three ways those may be processed separately and the resulting signals added to each other i or multiplied 11 or those may be processed in sequence by updating the signal at a target node each time an edge is processed iii The order in which the edges are processed in the last case may be adjusted by the user by defining rank based edge order see the User M anual for details 26 Pathway flow calculation Pathway flow calculation is the process of signal propagation from nodes at the first level to the nodes at the last level of the network At each level assigns weights to the edges according to the options set with Signal split rule and Split signal on button group no 5 and 6 in the Rules tab For each edge the signal at the edge is computed based on the edge type specific rule and the source and target node values and multiplied by its weight If a node has several incoming edges the edge signals are processed and accumulated in the target node according to Multiple signal processing rule and Signal processing order button group no 7 and 8 in the Rules tab To perform pathway flow calculation on the selected network after setting the respective options simply press the Calculate flow button Nodes and edge signals will be updated and the signals of all the nodes and edges w
4. are mapped to Cytoscape node and edge attribute tables and are also written in a backup file The following columns appear in the Node attribute table e psfc level This column appears after network sorting The level of each node is equal to the distance of that node to the pathway input The input nodes have level 0 the sink nodes have the highest levels in their branch psf_ x These columns are used for signal flow visualization and may be of little interest to the user The signal flow propagation is performed from 0 level nodes to higher level nodes After a level X iS processed a column psf lx appears where the node PSF signal values are kept For example in psf 12 only 0 1 and 2 levels are processed and the PSF signal values of nodes of those levels are updated while the rest of the nodes have PSF values equal to their original input values Note that when performing many iterations only values of the last iteration are displayed psf final This column contains the final PSF values of all the nodes when all the levels are processed In case of many iterations of PSF calculations the PSF values of the last iteration are displayed psfc_pval The significance p values of the PSF values for each node The following columns appear in the Edge attribute table psfc_isbackward This column appears after network sorting If true then the edge is a loop forming backward psfc _I x These columns are used for signal
5. continuous mapping of node signals onto the network The default is blue white red Node signal range the user may adjust the normal value of a node signal below which the signals are considered low and above which as high By default the middle signal is the average of the minimum and the maximum signals The user may set it as a constant Note that the middle value cannot be less than the minimum or more than the maximum Edge width range the maximum and minimum edge widths corresponding to the highest and the lowest edge signals for passthrough mapping to be applied on edges in the pathway The middle width can be adjusted within the min max range 17 14 15 16 17 18 Edge signal range the lowest middle and the highest signals of edges in the network The middle signal is the average between the lowest and the highest signals It can be adjusted within this range by the user PSFC log button Open the PSFC log file in the default text browser If the file is not opening open it manually from the path indicated in the Cytoscape Task M anager Save settings button saves the options from all the tabs Network label shows currently chosen network Calculate flow button Performs PSF calculation on the selected network 18 Options tab la Network Style Select PSFC Y Y General Options Rules Loops Help Algorithms Sorting algorithm 1 gt gt Topological sort So
6. convergence A convergence is achieved if the relative signal change between two consecutive iterations is less than 1 for all the nodes in the pathway If convergence is not achieved the algorithm will stop after 50 iterations 5 Calculate flow Backward edge gt o j interacti Y Y General Options Rules Loops Help Ignore feedback loops in the pathway J Iterate until convergence Convergence threshold 1 Yo Max number of iterations 50 The algorithm will iterate until reaching the convergence threshold at each node or iterating the max number of iterations See the user manual for details Precompute signals at loops The signal at Target nodes at feedback loops is precomputed and the rest of the algorithm proceeds as there vere no loops in the pathvay See the manual for details Now that the rules are set we can calculate the flow on the pathway Before that recall that all the nodes had fc values of 1 This means that all the nodes have the same value as the control state thus no signal will 12 be propagated in the network you may check this by calculating the en flow Thus we will change the fc value of the IGF1R node to 0 1 down regulating it and will calculate the signal propagation resulting from this perturbation The flow is calculated by clicking the Calculate flow button at the N bottom of the PSFC panel Set the fc value of IGF1R to 0 1
7. flow visualization and may be of little interest to the user The signal flow propagation is performed from 0 level nodes to higher level nodes After a level X is processed a column psf_Ix appears where the edge signals in the current state are kept For example in psf 12 only 0 1 and 2 levels are processed and the signals of the edges of those levels are updated while the rest of the edges have signals equal to 1 Note that when performing many iterations only values of the last iteration are displayed Getting started example the MAPK signaling pathway This is a quick demonstration of the app workflow The MAPK signaling pathway structure is taken from previous publications 1 Nelander S Wang W Nilsson B et al Models from experiments combinatorial drug perturbations of cancer cells Mol Syst Biol 2008 4 216 2 Feiglin A Hacohen A Sarusi A et al Static network structure can be used to model the phenotypic effects of perturbations in regulatory networks Bioinformatics 2012 28 21 2811 8 To follow the demo go to http big sci am apps psfc MAPK psfc configurations v2 rar and download the model network configuration files as well as the Excel output file for comparison of results 1 Load the network Go to File gt Import gt Network gt File and navigate to MAPK_model xml directory click Open to load the network into Cytoscape 4 Session New Session A IIA gt AAA File Edit Vie
8. mathematical function 4 Define flow propagation rules After step 2 3 each node is assigned a value and each edge is described with a specific behavior Flow propagation rules define i how the signal is passed from a source node to a target node via an edge and 11 how nodes receiving multiple signals from multiple source nodes process those signals and vice versa These rules are defined upon the user s choice 5 Define loop handling options Biological pathways contain positive and negative feedback loops In PSFC provides a number of options for handling loops in pathways 6 Significance calculation options The output of signal flow calculation is PSF values assigned to each node that describes the activity state of the node The significance of the PSF values is calculated with Bootstrap resampling The user may choose to compute significance values and may select different options for resamplings 7 Flow calculation After setting the configurations PSF computation is performed with a click of the Compute flow button PSFC outputs PSF values for all the nodes and edges in the pathway These values describe the amount of signal that a node carries and an edge transfers 8 Signal flow visualization The signal propagation may be visualized on the network via node color mapping and edge width mapping according to node and edge signal values 9 Final output of PSFC The PSF values of nodes and signals at edges
9. networkName SUIDnetworkSUID If the chosen network is in the current view the network label 15 is green otherwise it is red 2 Refresh button for Network combo box updates the list of currently loaded networks Should be 52_mapk_model_1 SUID346 z O E 2 sS v O Chek Dy o lt _7 8 aT I lt lt gt gt PI lt 10 o 3 4 5 Edge width 1 0 2 0 70 lt 13 Min Mid Max 0 29 0 77 1 25 gt 14 52_mapk_model_1 SUID346 _ __ 17 omar lt 18 used after a new network has been loaded or removed 3 Select Edge type attribute combo box for selecting the Edge type attribute from the selected network s Default Edge Table The Edge type attribute is used for defining the types of edges in the network and further setting edge type specific rules see Flow rule configurations section 4 Refresh button for Select Edge type attribute combo box updates the list of columns of the selected network s Default Edge Table Should be used when a column is added or removed from the table 10 11 12 13 Check button for Select Edge type attribute combo box shows the list of unique values in the selected Edge type column The list is just for checking all the unique types of edges in the network as well as for copying those for the rule configuration file see Flow rule configurations section Select Node data attribute combo box for selecting the Node data attribute f
10. 8010 COL PIE OMe een rrr mn nnn nant rere nnn renee Rate en nen nna ene nae eer ear 6 Getting started example the MAPK Signaling pathWay ooooccccccncnccccccininananana nono nononnnnnnnnnnnononinorana non rnnnnos 8 ie SACU TEEN NO 6 OCECOO E cee nee cere ents E 8 2 Load node and edge attributes cceescccccceessssssssceeeceeseceesseeeeeeeeseeeeeentateeeesseseeseaaetteesesssseenntanees 9 3 SESION PO DADA OA TUS ar AA AAA AA EA 10 A Setloop handling ODTONS c sesssusaananacasaeaeranenanenae onnantannsanakanstaaataastaneta onaeatanatanakanstaaatanrbanenaneneeenidad 11 PP cons qenceaneeansnansaunsmamestanaancnanteesetuaenenateast 12 Ox RIF OlcId0U cern r nT rT ttn rr o 13 Te Visualize te TNO Wasa erreurs eer ead coa 14 1G 66 ag 0 0 aia a ara a nen nent pene nen penn mene nrne nn Rene O en nen nen nen nen nee ene nnn eee eee eee 16 GENE Falr EG TE T T A T 16 OBUCAS added 19 PAU css eee nn E eer eRe ne enn E AE eee ce A E er eer ee ee eee 21 Po er on E o IA 24 NEW OTK SOFLING n sctncancasncasaarannenanetnbatanasaanaanasa gage sasamnnnnutniuannacaaatanasa gaan aatunneenutabuaigagntatanadaaaaeaaqunonenemnrnenast 24 FIOW DFO PaCallON TUS aaa atada 25 Pathway flow calculation ooooccccnnnccccicnnnnncananono nono nonnnnnnnononononono nano rro rr nr rr rn rr r rr rr rr rr ran ann nr rr rn rrnrrnnnnnnnss 21 POWA A TON NE eee rer re ee 21 Introduction The Pathway Signal Flow Calculator PSFC is a Cytoscape app for
11. ID 11 name p RAF type activation weight 0 29 rank 0 loopCount 0 Edge source ID 0 name PKCdelta target ID 6 name Gi arrest type inhibition weight 0 33 rank 0 loopCount 0 Edge source ID 1 name IGFiR target ID 6 name Gi arrest type inhibition weight 0 03 rank 0 loopCount 0 Edge source ID 3 name p P70S6K target ID 2 name p S6 type activation weight 1 0 rank 0 loopCount 0 Additionally the sorting level wise PSF values of the nodes after the last iteration are stored in an xls file with the name of the network The file is located in the Cytoscape directory usually in the c Users User CytoscapeConfiguration app data PSFC path 7 Visualize the flow The flow propagation on the last iteration is visualized based on level wise PSF values of nodes The signal is visualized from lower levels of the pathway to the highest ones Note that the layout of the pathway does not correspond to the sorted order as shown in step 4 as the option for changing the network layout was not chosen The flow propagation is visualized based on the signal passing from the inputs of the network to its outputs from lower levels to higher In case of multiple iterations the last iteration flow is shown To see the original node values simply hit the home button among the flow control buttons or slide to level 0 The final state of the signals on the pathway is at the last level
12. Note that a number of columns will appear in the node and edge attribute tables that describe the PSF values of nodes and edges during signal propagation on the network from lower levels input nodes to higher levels sink nodes There is a column named psf final which keeps the final output of PSFC Node data attribute mo raan mTOR apoptosis fc v o Flow visualization GFIR Levek 6 Table Panel amp D ee oo 3 wW f z Show state Play flow Shared name isExclu Label p MEK p MEK false EGFR EGFR false EGFR 1 0 p RAF p RAF false p RAF 1 0 1 0 1 0 p eIF4E p elFAE false p eIF4E 1 0 1 0 1 0 PI3K PI3K false PI3K 1 0 1 0 1 0 p AKT p AKT false p AKT 1 0 1 0 1 0 PSFC log 52_mapk_model_1 SUID52 n FRK n FRK fake n FRK 10 10 1 0 4 Save settings Node Table Edge Table o Tae Final PSF values of the nodes 6 PSFC output Besides the signal values in the node and edge attribute tables PSFC tracks a log file where the input the option and the results of the computation are stored The log file may be accessed by clicking the PSFC log button at the bottom of the PSFC panel According to the results of our calculations convergence was reached at iteration 7 The psf signal results can be found in the MAPK configurations v2 rar the file MAPK_ scores output xlsx 13 Iteration 5
13. USER 2SFC 1 0 2 MANUAL Lilit Nersisyan and Arsen Arakelyan Institute of Molecular Biology NAS Armenia Graham Johnson Megan Riel M ehan Alexander Pico UCSF and Gladstone Institutes USA November 2015 PSFC User Manual Table of Contents SEC US Mist 1 A nese nee N ee ee ee ee ee ee errr 3 DEVEIODINEME sacrassasaassasqanensqenetetnantaecnatagacaaageaanennnenetenedumig aanaaacagagesaaannnenenenetanam aaaunanagagaasaasoneneaunemnmaneaaaananagage 3 Ciara aaa 3 Nersisyan L Johnson G Riel M ehan M et al PSFC a Pathway Signal Flow Calculator App for Cytoscape version 1 referees 1 approved Fl000Research 2015 4 480 dol 10 12688 f1000research 6706 1 3 PN Pe o EE TE EE EEEE EEEO EE EEE 3 Gettmo PP O E 0 EREE A OEE EEE EAE EA EE REAG 4 a rn E E O E ee er eer EEE EE E EE 4 EMEI ES E e A EA NO 5 be LOOTE NE NOT ea E O rere 5 2s load Node MPU dold AAA P gt 5 34 Loadedge types QUEL OIL E ere N a 6 4 Define flow propagation rules coononoccccooonnccnonoconononononononananon ron nn non onrnnnrrrn rro nnronnnano rr rn rn rr rannnnnnnos 6 de DENNIS ele op nanding OPONE 6 6 Significance calculation OPTIONS cooococccccnccccnononannonnncncnnnnnn non nnnnnnononnnononnrnnrrnnnnnnor rr rnnnrrrnnnnnrrnnennss 6 Te HOWE cloaca 6 8 Signal flow visualization ccccccccinoonnnnononrnconononononnnnnonononononanon non r or rr nro nr nn rrrrr nro roronnnnno nn nor rr rn rnnnnnnnnos 6 e EIN 101
14. al Public License version 3 The license can be found at http www gnu org licenses gpl html Getting started PSFC has been tested to work with Cytoscape version 3 2 0 and higher on 64 bit systems Running it on other systems or with previous versions of Cytoscape may result in errors Installation In Cytoscape go to Apps gt App Manager choose PSFC under and click on the install button In case of successful installation the PSFC app tab should appear in the west panel of Cytoscape Figure 1 amp Session New Session a AT e File Edit View Select Layout Apps Tools Help DAA ee Pathway Scoring Control Panel lg Network Style 3 WV aa General Opti Rules L H i AO Network and attributes Network all apps 92 MCODE apps by tag MetDisease 1 0 0 Edge type attribute MetScape y 0 Check ModuLand 2 0 pS PC ncINetView Node data attribute PanGIA Pathway Signal Flow Calculator r PathExplorer Installed o Pathway Scoring Application Flow visualization Leve 50 I stat ODI NAT A A A AT LAP OD i Table Panel Rene Play fiow e D ez oc PA Install from File View on App Store In Close PSFC log ne chosen Save settings Enter search term 9 Node Table Edge Table Network labi Memory OK O Figure 1 Installation of PSFC General use case General Opt
15. alues for each node is used to calculate the z score and the p value see PSFC algorithms for details The p values will appear in the psfc_pval node attribute column Radio buttons specifying the mode of Bootstrap resampling a Sample centric the values of the nodes in the network will be reshuffled among each other during resampling b Gene centric the value of each node is randomly chosen from a supplied distribution of node values Such a distribution e g may come from microarray chips where the expression of a single gene is measured in several samples Exp M atrix button for Gene centric option select the file where the value distributions of nodes are specified This is a tab delimited file where the first column contains the node names and the tab delimited values of the node are written in each row 20 Rules tab fg Network Style Select PSFC Y Y Y General Options Rules Loops Help Open Rule Presets Guide 1 2 Simple rules ary EdgeTypes config file Rule config file 3 gt Choose file e 4 edgeTypeRuleNam ruleNameRule co 5 Multiple input and output edge rules a 6 gt 3 Signal split rule 6a None Proportional a Supplied weights O G 6b Y gt Split signal on Incoming edges Outgoing edges 3 gt Multiple signal processing rule Updated node scores Multiplication Addition 94 o gt Signa processing or
16. calculation of pathway signal flow based On input data and pathway topology The app provides a flexible interface for setting flow propagation rules and applying algorithms of wide range to assess activity states of pathway nodes PSF signals The app supports visualization of signal propagation on the network Development The app has been developed by Lilit Nersisyan and Arsen Arakelyan from the Bioinformatics Group at the Institute of Molecular Biology of the National Academy of Sciences of the Republic of Armenia IMB NAS RA in collaboration with Graham Johnson and Megan Riel M ehan from University of California San Francisco and Alexander Pico from Gladstone Institutes The project was funded by Google Summer of Code 2014 and Armenian National Science and Educational Foundation Grant research grant molbio 3818 PI LN Citation When using the app in your research please link to the app s webpage at http apps cytoscape org apps psfc and cite our paper as Nersisyan L Johnson G Riel Mehan M et al PSFC a Pathway Signal Flow Calculator App for Cytoscape version 1 referees 1 approved Fl000Research 2015 4 480 doi 10 12688 f1000research 6706 1 License Copyright O 2015 Lilit Nersisyan IMB NAS RA Arsen Arakelyan IMB NAS RA Graham Johnson UCSF M egan Riel M ehan UCSF Alexander Pico Gladstone Institutes This program is free software you can redistribute it and or modify it under the terms of the GNU Gener
17. der o None Edge ranks lt 1 9b PSFC log 52_mapk_model_1 SUID52 Save settings The Rules tab contains the following components 1 Open Rule Presets Guide button opens the PDF file where preset options for the Rules tab are explained If the PDF doesn t open for some reason you should open manually from the PSFC app data directory specified in the task monitor 2 Simple rules panel for setting edge type specific rules onto single edges Defines how a signal is transferred from a source to a target node via a single edge 3 Choose EdgeTypes config file button sets the configuration file containing edge type rule name mapping This should be a tab delimited file where the first column contains edge types and the second column contains the names of the functions assigned to each edge type 4 Choose Rule config file button sets the configuration file containing function name function mapping This should be a tab delimited file where the first column contains the function names and the second column contains the function specifications corresponding to each name The 21 functions should be of the form f source target where source Is the source node signal and target is the target node value Multiple input and output edge rules panel for setting rules to process signals from multiple edges starting from a single source node or ending on a single target node Signal split rule specifies whether to split mul
18. eedback loops will not be considered during PSF calculations 2 lterate until convergence checkbox if checked the PSF calculations will consider loop forming backward edges as other edges in the pathway and the algorithm will run for several rounds until reaching convergence Convergence in reached if the relative percent difference between node signals in two consecutive iterations Is less than the specified threshold 2a If convergence is not reached the computation will finish after the specified maximum number of iterations 2b 3 Precompute signals at loops checkbox if checked the signal at backward edges Is firstly computed and their target node values are updated Afterwards the PSF calculation is run on the whole network in Ignore feedback loops mode 23 PSFC algorithms Network sorting For computation of pathway signal flow propagation the network should initially be sorted based on distances of intermediate nodes from the input nodes where input nodes i e nodes that don t have incoming edges should be assigned level 0 In order to see network sorting results before pathway flow calculation the user may press the Sort button 4 in Network tab panel PSFC will sort the selected network and optionally apply the level based layout onto network view psfc level column will appear in the default node table of the selected network where the level of each node will be indicated Node levels after network sorting
19. essed one after the other in random order alternatively the user may supply Edge ranks defining the relative priorities of the edges the smaller the rank the sooner the edge will be processed 8a The CyColumn in Default Edge table of the selected network which contains edge rank attribute values should be of Floating Point type is active when Edge ranks option is selected 8b Refresh button for edge ranks column updates the list of columns of the selected network s Default Edge Table should be used when a column is added or removed from the table 22 Loops tab la Network Style Select PSFC Y Y General Options Rules Loops Help 1 gt Ignore feedback loops in the pathway 2 gt gt _ Iterate until convergence Convergence threshold 1 la Max number of iterations 50 lt _ lt 2b The algorithm vill iterate until reaching the convergence threshold at each node or iterating the max number of iterations See the user manual for details 5 gt gt Precompute signals at loops The signal at Target nodes at feedback loops is precomputed and the rest of the algorithm proceeds as there vere no loops in the pathvay See the manual for details PSFC log 52_mapk_model_1 SUID52 Save settings The Loops tab specifies how to handle feedback loops in a pathway It has the following three options 1 Ignore feedback loops in the pathway checkbox if checked the backward edges that form f
20. ill be visible in newly created or updated columns in default node and edge tables of the selected network in Cytoscape The column names will have the suffix psf followed by the level value Signal values are also kept in the score backup xls file kept in the apps directory usually located at C Users User CytoscapeConfiguration app data PSFC the address is indicated in TaskM onitor The log of PSF calculation steps are debugged in PSFC log file which is located in the apps directory usually located at C Users User CytoscapeConfiguration app data PSFC PSFC log and can be opened by pressing the PSFC log button in the bottom of the panel Flow visualization The Flow visualization panel allows the user to see the updated node signals by applying color based visual mapping of the values in the psf 1 columns The Contorl buttons allow for visualizing the network at different levels or time steps which will show the node signals at the state when the flow propagation has reached the nodes of the level indicated on the levels slider or level textbox In case of iterative loop handling the pathway flow is visualized for the last iteration only To see the original node values are visualized at level 0 which is reached by the home button among the flow control buttons or with the slider The final state of the signals on the pathway Is at the last level or time step To see the last state hit the end control but
21. ions Rules Loops Help General Options Rules Loops Help Open Rule Presets Guide Network and attributes Network Simple rules EdgeTypes config file Rule config file 52_mapk_model_1 SUIDS2 y o Edge type attribute AA FESE Load the network with node and type O check edge attributes Node data attribute o Multiple input and output edge rules exp Yv Signal split rule E None Proportional Equal Flow visualization Supplied weights Da Set up signal propagation rules Set up significance calculation i Incoming edges Outgoing 7 on the network i Showstate General Options Rules Loops Help A l Ignore feedback loops in the pathway Algorithms Sorting algorithm Calculate flow v Iterate until convergence Topological sort X Sort Convergence threshold 1 Yo _ Change network layout after sorting Max number of iterations 50 Perform PSF ca Icu lation The algorithm will iterate until reaching Significance calculation the convergence threshold at each node or iterating the max number of iterations Calculate significance after calculating flow See the user manual for details Sampling type o Sample centric a P oz Visualize signal propagation Gene centric Expr Matrix The signal at Target nodes at feedback No of samplings 200 loops is precomputed and the rest of the Figure 2 The general use case of PSFC 1 Load the network PSFC work
22. on and activation state However if one type of molecule has several interacting partners those may compete with each other and the interaction capacity of the molecule may be split between those partners Following this idea PSFC is implemented with different options for splitting the signal among multiple outgoing or incoming edges The splitting of the signals is performed as follows 25 i Edges are assigned weights according to the splitting rule If the rule is set to Equal the weight of each edge Is calculated by dividing 1 to the number of edges If the rule is set to Proportional and edges are split on edges outgoing from a single source node each edge gets a weight equal to the relative signals of its target nodes If edges ending at a single target are split with the Proportional rule the relative source signals are used for weight assignment Alternatively the user may supply weights of their choice ii The assigned edge weights are multiplied by the source node signal if splitting is performed on multiple outgoing edges or by the target node value in case of splitting on multiple incoming edges ili Finally the edge associated function is applied to the weighted source and target node signals Comparison of splitting options 3 w edge weight after splitting Split on s source node signal incoming t target node signal vs rule s t outgoing gt a Split equal
23. or time step To see the last state hit the end control button or slide to the very end of the level slider The intermediate states are visualized with forward and reverse controls 14 a gt Level 3 Level Timestep 6 iN 5 2 3 4 5 e Cobe scheme tage wetth aa 10 40 76 Mn M Mae Mn Md Max 10 10 10 0 0 os 10 li Level 0 i mfe Ps Level 5 e y pas l Level 2 Ps li Level 6 a E ll E Note that the nodes are colored from blue to red for signals from low to high The median point is adjusted to be equal to 1 as it corresponds to the normal state The middle signal for edges is also chosen to be 1 and the width of edges corresponds to the signal intensity The colors middle node signal value edge width range and middle edge signal value are adjustable by the user See the section GUI components for details 15 GUI components General tab fe Network Style Select PSFC Y General Options Rules Loops Help Network and attributes Network 1 gt Edge type attribute 3 gt we Node data attribute 6 gt Flow visualization 9 Level Timestep 1 Kl Y 0 1 2 Color scheme 1 gt Y B Min Mid Max 12 Y 01 10 1 39 15 gt 16 gt PSFC log Save settings Contains the following components 1 Network combo box for selection of the network under interest The networks are represented in the form
24. rom selected network s Default Node Table The Node data attribute is used for setting the initial values of nodes for flow calculation Node data values can be either row of logarithmic expression values fold change or log fold change of expression values etc Refresh button for Select Node data attribute combo box updates the list of columns of the selected network s Default Node Table Should be used when a column is added or removed from the table Flow visualization panel for performing visual mapping of node and edge signals after SPF calculation Level textbox and slider for setting the PSF state at the specified level timestep in the sorted network A level of the sorted network specifies the distance of nodes from the pathway input The signal is propagated from the 0 level to the highest level in the pathway see Network sorting for details Control buttons performs color gradient visual mapping based on the node signals and edge width mapping based on the edge signals at the chosen level There is the home button which shows the signals at level 0 also corresponding to original values of nodes and the end button which shows the last level or final signals The intermediate states can be viewed with the forward and reverse buttons as well as the slider Note that in case of many iterations the signals of the last iteration only are shown Color scheme for node signals the user may adjust the color key to be used in
25. s or many edges are incoming to the same target node These rules are set with the Multiple input and output edge rules panel in the Rules tab For details on various rules in the panel refer to further sections of the manual We will now discuss the rules set for the M APK signaling pathway 10 e The Split rule is set to Proportional and the Split v Y signal on is set to Incoming This means that the General options Rules Lonas Hee signal coming from multiple edges to the same Open Rule Presets Guide target node will be split or assigned weights based on relative values of their source nodes Biologically this simulates the fact that when interacting with multiple competing partners the interaction intensity of the molecule depends on relative concentrations activity states of those partners The Multiple signal processing rule is set to Addition With this rule the signals from multiple edges coming to the same target node will be Simple rules EdgeTypes config file Rule config file l Choose file Choose file edgeTypeRuleNam ruleNameRule co Multiple input and output edge rules Signal split rule None Proportional Equal Supplied weights Split signal on Incoming edges Outgoing edges Multiple signal processing rule Updated node scores Multiplication Addition Signal processing order None Edge ranks added to each other and the sum will be the signal tran
26. s on any pathway network loaded into Cytoscape environment The network may either be loaded with File gt Import gt Network from a local file or from the web with available pathway parser apps such as WikiPathways CyKEGGParser KEGGScape ReactomeFlPlugin etc Instructions on how to load create and edit networks can be found at Cytoscape User Manual page http wiki cytoscape org Cytoscape 3 UserM anual Cytoscape 3 2BAC8 UserM anual 2BAC8 Quick Tour Welcome Screen 2 Load node input data The nature of node data may vary it can be absolute or relative expression of genes in each node fold change of expression values node ranks protein concentration protein phosphorylation level etc Basically this is any data related to the activity states or the quantities of the nodes The node data should be loaded in Cytoscape node attribute table This can be done manually in Cytoscape Table Panel via File gt Import gt Table from a local file For more details on how to work with node attributes the user is referred to Cytoscape User M anual page http wiki cytoscape org Cytoscape 3 UserM anual Niode and Edge Column Data 3 Load edge types attribute The edge types actually define how the signal is propagated between nodes in the network The type of the edge should be loaded in Cytoscape edge attribute table similar to node attributes as described above As explained below each edge type may be associated with a different
27. sferred to the target edge 4 Set loop handling options Biological pathways contain positive and negative feedback loops which are important regulators of pathway activity and properties The MAPK signaling pathway contains many negative feedback loops Those can be seen by applying the Topological sorting algorithm by clicking the Sort button in the Options tab and checking the Change network layout after sorting checkbox 11 Ox 52 mapk_model_1 Control Panel la Network Style Select PSFC Y Y General Options Rules Loops Help Algorithms Sorting algorithm Topological sort J Change network layout after N Significance calculation _ Calculate significance after calculating flow Sampling type Sample centri Gene centri Expr Matrix No of samplings 200 Table Panel smpe co amp g shared shared name 97 97 98 98 PSFC log 52_mapk_model_1 SUID187 94 94 Save settings Node Table Edge Table Network Table The backward edges introduce cycles in the graph complicate signal flow propagation on the pathway Among the three choices for loop handling presented in the Loops tab we will choose the with its threshold 1 Maximum number of iterations 50 This means Iterate until convergence option parameters Convergence that the flow propagation will be performed on the whole pathway for multiple times updating the signal flow on the nodes after each iteration until reaching
28. t e 2 3 gt Change network layout after sorting Significance calculation e Calculate significance after calculating flow Sampling type 5 Sample centric Gene centric Expr Matrix AAA No of samplings 200 lt _ 7 PSFC log 52_mapk_model_1 SUID52 Save settings Contains the following components 1 Sorting algorithm combo box for selection of the sorting algorithm to be applied on the network Currently only a modified version of topological sort algorithm is supported see PSFC algorithms for details In short sorting assigns each node a level which indicates the distance of that node from the nearest input node The signal flow is propagated from the input nodes that have a level of 0 to the sink nodes that have highest level values Sort button for applying selected sorting algorithm onto selected network and updating the network layout respectively Change network layout after sorting checkbox if checked the network layout will be changed by arranging the nodes from left to right from lower to higher levels Note that the original layout will not be preserved Calculate Significance after calculating flow checkbox if checked significance values of PSF values for all the nodes will be computed by Bootstrap resampling The node input values are reshuffled during each resampling and the PSF values are computed again After multiple resamplings the 19 obtained distribution of PSF v
29. te from its source node to the target Rules for signal flow propagation for single edges are defined with two files e EdgeTypes config file should be a tab delimited file where the first column indicates the edge types and the second column indicates rule names corresponding to each edge type All the edge types in the selected network should be present in the config file in order for PSF calculation to be performed e Rule config file should be a tab delimited file where the first column indicates the rule names and the second column indicates the functions corresponding to each rule name All rule names corresponding to all the edge types in the selected network should be present in this file Functions are defined as mathematical equations of the form f source target with variables named source and target for source and target nodes respectively The user defined edge type functions are parsed with Exp4j Java library for symbolic operations The parsable functions include a number of mathematical operations including sin cos log natural logarithm and constants such as Math Pl Math E Rules for multiple incoming and outgoing signals are applied for the cases when multiple edges go out of a single source node multiple outgoing edges or multiple edges go to a single target node multiple incoming edges Signal splitting Generally intensity of interactions between molecules largely depends on their concentrati
30. tiple edges outgoing from a single source or incoming to a single target node i by dividing to the number of edges Equal 11 by assigning weights to edges proportional to relative values of the target nodes or relative signals of the source nodes Proportional 111 by assigning the edges user supplied weights Supplied Weights iv alternatively there can be no splitting at all None 6a The CyColumn in Default Edge table of the selected network which contains edge weight attribute values should be of Floating Point type is active when Supplied weights option Is selected 6b Refresh button for supplied weights column updates the list of columns of the selected network s Default Edge Table should be used when a column is added or removed from the table Split signal on specifies which edges to apply splitting on Incoming multiple edges ending on a single target node Outgoing multiple edges starting from a single source node Multiple signal processing rule specifies how multiple signals received by a single node should be processed be added Addition or multiplied M ultiplication and replace the previous node signal or the signals should be processed one after another each of them replacing the previous node signal Updated node scores Signal processing order if the Multiple signal processing rule is set to Update node scores the order in which the edges are processed matters If set to None the edges will be proc
31. ton or slide to the very end of the level slider The intermediate states are visualized with forward and reverse controls Note that the nodes are colored from blue to red for signals from low to high The median point is adjusted to be equal to 1 as it corresponds to the normal state The middle signal for edges is also chosen to be 1 and the width of edges corresponds to the signal intensity The colors middle node signal value edge width range and middle edge signal value are adjustable by the user See the section GUI components for details 21 LSF Pathway Signal Flow Calculator 2015 Lilit Nersisyan IM B NAS RA Arsen Arakelyan IM B NAS RA Graham Johnson UCSF M egan Riel M ehan UCSF Alexander Pico Gladstone Institutes Licensed under GNU General Public License version 3 Discussion group https groups google com forum forum psfc discussion group Correspondence Lilit Nersisyan _nersisyan mb sci am Group of Bioinformatics Institute of M olecular Biology NAS RA 7 Hasratyan str 0014 Yerevan Armenia November 2015 28
32. ules In biological pathway representations an edge is a connector between a source and a target node which transfers the signal from the source to the target Edges may have different functional annotations In our network there are edges of type activation that increase the activity of the target node and those of type inhibition that decrease it We would like to simulate these functional annotations of edges by defining a function for each of those For this we will go the Rules tab Simple rules panel There are two configuration files the Edge types config file is a tab separated file where each edge type is assigned a function name The mathematical functions corresponding to each of these function names are defined in the Rule config file Those are in the form f source target where source and target variables are source signal and target value respectively fg Network Style Select PSFC Y y General Options Rules Loops Help Open Rule Presets Guide Simple rules EdgeTypes config file Rule config file a Encoding Help 100 Lister D Dropbox File Edit Options source target 1 source target File Edit Options Encoding Help 100 activation a Function names Functions Multiple input and output edge rules Following these simple rules for single edges we should also define how the signal should be transferred if a single source node has multiple outgoing edge
33. w Select Layout Apps Tools Help Recent Session New gt Open Ctrl O Ox Save Ctri S Save As Ctri Shift S Import Network Export i Table Style Ontology and Annotation Run Seison New Seison He tdt View Select Layout Apps Took Help w ot OH a aa 080 GS ta ts oe Control Panel ox 2 SE map made 7 gt PP lt lt TEE ly Network Style Select PSC General Options Rules Loops Help Network and attributes Press on the refresh buttons in the PSFC General tab to notify it about the newly loaded network and attributes From the Network dropdown menu choose the M APK network _ Tg Network Style Select PSFC h General Options Rules Loops Help Network and attributes Network Edge type attribute Node data attribute 2 Load node and edge attributes The MAPK network contains two types of edges activation and inhibition In the following steps each of these edge types will be assigned a different mathematical function for signal transfer The edge type attribute is loaded in the Cytoscape edge I N attribute table in the type column In PSFC General tab select the type column from Control Panel Ox 52 mapk model 1 lg Network Style Select PSFC Y v General options Rules Loops Help Network and attributes Network 52_mapk_model_1 SUIDS2 Edge type attribute type Node data attribute Show state Table Panel Pla
34. y flow o D es owo D Mm f z Y h General Options Rules Loops Help Network and attributes Network 52_mapk_model_1 SUID52 x 52_mapk_model_1 SUID52 SUID v Check Node data attribute SUID d activation Check smh ihr inhibition Sy zx j m Me 1 Tatte Panel o D rr o 8 Ja shares shares PITA eenaa toe 97 oF activation os gt T octetos S2_maph model 1509052 os s Ston s E Jamnan mode Tabie f pe Tabie Network Table SSS shared name p MEK p MEK EGFR EGFR false EGFR false p MEK PSFC log 52_mapk_model_1 SUID52 p RAF p RAF E i false p RAF ae p eIF4E p eIF4E false p eIF4E z aiculate wW Save settings Node Table Edge Table Network Table the Edge type attribute drop down menu The node data is in the node attribute table under the column fe The values in this column represent fold changes of protein phosphorylation levels in a studied condition compared to the control In our network all the nodes have initial values of 1 0 that correspond to the control not changed state Furthermore we will change the values of different nodes of the pathway to see how these changes propagate to other nodes and to the sink nodes of the pathway From the Node data attribute drop down menu in the PSFC General tab choose the fc column 3 Set signal propagation rules Simple r
Download Pdf Manuals
Related Search