Home

VoICE Manual

1. Enter an animal ID for the similarity batch here I will use example Once a valid ID has been entered the yellow status bar will change to green and say Ready Press Score Similarity The yellow status bar will turn red while similarity is scored Running then yellow again when complete Done Awaiting next animal 9 lt Student Versi lt Student Versi VoICt VoICt Similarity Scorina m Similarity Scoring Select a folder Select a folder mon REE conn ES Le S Enter Animal ID example Enter Animal ID example Score Similarity Score Similarity VoICE m Similarity Scoring Select a folder cote romena ERRE anon Enter Animal ID example Score Similarity 30 Step 2 Assign Clustered Syllables to Canonical Call Types After scoring similarity the folder in the Current Directory field of the Similarity Scoring panel will contain a CSV file which is the result of the similarity batch Select this file in both the Animal 1 and Animal 2 panels by pressing the Select a Similarity Batch button in each Important Note If the user wishes to be blind to the animal s genotype when assigning calls select similarity batches from animals of different genotypes in Animal 1 and Animal 2 panels 9 lt Student Version gt voice_usv VoICE USV m Similarity Scoring Assianment
2. 23 Depending on the GS Threshold selected the VoICE pipeline will branch in a number of directions In order to illustrate all possible options for this tutorial will use a GS threshold 50 that will allow for each module to be demonstrated To continue with the tutorial enter a GS threshold of 50 and press Assign Syllables The following pages will illustrate the different modules that can launch 24 Step 3 Navigate Through Modules The Tiebreaking Module This module launches when one or more syllables were determined to need a manual tiebreak The user will proceed through these syllables one at a time in this module For each syllable the module displays the syllable in question within the context of a motif underlined in red Below spectrograms of a representative from each cluster are shown The representative spectrograms are ordered from left to right in GS descending order 9 lt Student Version gt tiebreaking_module ON ey VoICE Finch Syllable Tiebreaking Module Syllables that passed your global similarity threshold with multiple clusters and did not show a statistically significant difference between those clusters must be manually tiebroken The syllable in question is underlined in red top and a representative from each prospective cluster displayed in order of highest to lowest global similarity are shown below Use the zoom tool for greater detail mo ie i VO SE so 0 COS D a baa Rn s
3. Select Syllable Table button and navigate to the sample_data folder then select the Yellow119_ 122013 xls file and hit Open The Similarity Batch Module will now display the file path to the selected syllable table 9 lt Student Version gt similarity_module Similarity Batch Module Current Syllable Tabie SA Similarity Scorina Clustering Cut WAV Files Run Similarity Batch Cluster Syllables Min Dur 6 i Determine Merging Threshold Win Size 41 ee ee Status Updates Will Display Here Once the Current Syllable Table field is populated with information press Cut WAV Files The button clicked will turn yellow while running and then green when done You will then see a new folder entitled cut_wavs in the directory containing your data oo lt Studi LA lt Studi Sim Sim Current Syllable Table Volumes Ma Current Syllable Table Volumss Mal Sim Sim E E Status Status Cutting WAV Once the Cut WAV Files button has turned green you may begin running the Similarity Batch by clicking the Run Similarity Batch button Note The similarity batch code has been optimized to mimic the settings in SAP s feature batch as closely as possible The similarity batch will begin running A progress bar will spawn in the MATLAB desktop module Note A properly installed and configured P
4. Launch R as an Administrator then 1 Install the GO db package source http www bioconductor org biocLite R biocLite GO db 2 Install the WGCNA package install packages WGCNA 3 Install the gdata package install packages gdata 4 Install the impute and preprocessCore packages from bioconductor source http www bioconductor org biocLite R biocLite impute biocLite preprocessCore 5 Install the ggmap and png packages install packages ggmap install packages png Install SoX http sox sourceforge net then add the install directory to the system Path variable see Path modification instructions for R installation above Launch MATLAB then type voice at the command line to launch the GUI VoICE USV Windows Installation Install MATLAB R2014a with Signal Processing Toolbox Unzip VoICE_usv zip to its own directory and add this directory to the MATLAB Path Install R 3 1 2 http r project org and add the bin folder in the install directory to the system Path variable See https www java com en download help path xml for details specific to your Windows version regarding modification of the Windows path Launch R as an Administrator then 1 Install the GO db package source http www bioconductor org biocLite R biocLite GO db 2 Install the WGCNA package install packages WGCNA 3 Install the i
5. then want to compare a second day s recordings to those clusters Starting material 1 The sample data directory following the conclusion of the previous tutorial Think of this as Day 1 2 A copy of the original sample __data directory before starting the previous tutorial This directory will be referred to as sample_data 2 Think of this as Day 2 We expect a perfect match between Day 1 and Day 2 since they are actually the same recordings D sample_data H Em iol By HY 3 Q Shared Folder Name Size Kind syntax_summary csv 553 bytes comm values 3 workspace Rdata 465 KB R Document pnext csv 26 bytes comm values cluster_dendrogram pdf 14 KB PDF Document gt cluster_tables Folder gt D cluster_tables_mat Folder gt D joined_clusters Folder gt D sorted_syllables Folder unusedColors txt 7 KB text 1 colnames txt 254 bytes text igsdata csv 961 bytes comm values tree_trim_curve pdf 4 KB PDF Document tree_trim_curve png 21 KB PNG image igs Rdata 450KB R Document similarity_batch_self csv 10 MB comm values similarity_batch_completed mat 19 2MB MATLAB Data sample_data 2 acoustic_data csv 75KB comm values CO tous E ie mv 33 gt E cut_syllables Folder Shared Folder M Yellow119_122013 xis 114 KB Micros ksheet bade ae Yellow119_Decemb 013_29836554 wav 841 KB Waveform audio P Vellow119 122013 xis aan Yellow119_Decemb
6. 013_29828574 wav 625 KB Waveform audio Yellow119 Decemb 013 29836554 wav 841 KB Yellow119_Decemb 013_29818694 wav 552 KB Waveform audio gt Yellow119_Decemb 013_29828574 wav 625 KB Yellow119_Decemb 013_29799624 wav 1 5 MB Waveform audio Yellow119_Decemb 013_29818694 wav 552 KB Yellow119_Decemb 013_29779153 wav 1 8MB Waveform audio gt Yellow119_Decemb 013_29799624 wav 1 5 MB Yellow119_Decemb 013_29765924 wav 1 2MB Waveform audio OE A O rsdn se gt Yellow119_Decemb 013_29765924 wav 1 2 MB Yellow119_Decemb 013_29745682 wav 1 8 MB Waveform audio aar iaai TA ai pre gt Yellow119_Decemb 013_29705993 wav 1 8 MB Waveform audio Yellow119 Decemb 013 29705993 wav 1 8 MB Yellow119_Decemb 013_29691843 wav 1 3 MB Waveform audio Yellow119_Decemb 013_29691843 wav 1 3 MB gt Yellow119_Decemb 013_29684562 wav 413 KB Waveform audio Yellow119_Decemb 013_29684562 wav 413 KB Yellow119_Decemb 013_29664384 wav 1 6MB Waveform audio __ Yellow119_Decemb 013_29664384 wav 1 6 MB 19 Launch VoICE by typing voice at the MATLAB command line Click Compare Two Recording Sessions eooo lt Student Version gt voice VoICE Finch Cluster a Single Recording Session Compare Two Recording Sessions This will launch the assignment module E ES lt Student Version gt assignment_module VoICE Finch Assignment Module Syllables from the assignment directory will be compared to
7. Animal 1 Select a Similarity Batch Select a folder Assignment Animal 2 Select a Similarity Batch Enter Animal ID example Score Similarity Assianment Options ed Cohesion Threshold 0 80 Min Cluster Size 5 Launch Assignment Module Once similarity batches are loaded the user can optionally edit the Cohesion Threshold and Min Cluster Size fields in the Assignment Options panel Cohesion Threshold Clusters must display an average level of correlation with their eigencall at or above this threshold in order to be assigned by inspection of the call most like the eigencall Otherwise calls within the cluster are assigned individually Min Cluster Size Clusters must be at least this large in order to be considered for classification by inspection of the call most like the eigencall Otherwise calls within the cluster are assigned individually For the purposes of this tutorial we are analyzing only 20 total calls Thus will decrease the minimum cluster size to 2 and proceed by pressing Launch Assignment Module 31 The Assignment Module lt Student Version gt voice_usv_assign gt o c D 2 T The assignment module is self explanatory Based on your selections in the previous window the total 0 005 0 01 0 015 0 02 0 025 0 03 0 035 0 04 0 045 Time Working On File 1 Back number of calls that the user must assign will be high
8. into the module only a single syllable type was discovered Thus the choice for merging threshold is simple Choose the merging threshold then hit Generate Clusters When the button turns green hit Reassign Syllables This will launch the Reassignment Module 26 The Reassignment Module within the assignment pipeline The reassignment module launches and appears similar to the one discussed in the Clustering a Single Recording Session tutorial See page 16 Our novel cluster from the previous step is now present Since we know these syllables belong in a different cluster explore the reassignment tools to place them in their proper cluster When complete hit No More Changes Done Once the reassignment module is complete new items will appear in the sample_data 2 folder These include syntax_summary_assign_ folderlD ref_ folderlID csv A comma separated file containing transition probability matrices and syntax entropy scores for the assignment session and the reference session Also present in this document are syntax similarity scores and frequencies of occurrence for each syllable type in each recording session cluster_tables_ assigned A folder containing csv files named by cluster with the acoustic data for each syllable in the cluster joined_clusters_assigned A folder containing joined wav files of all the syllables in each cluster sorted_syllables_assigned A folder containing
9. number Time j i i Delete Syllabies r E EER ee A HSR ASAA oT 4 voly f a_n 3 ae ve he ni Pre ah as ae LE fils ER P x j RSi SN aS s AR IAI ES 144 j i f i N W i Dahan teat qe f i SNE ee oa rl PAAR IS H RAURA TE MOAS UESAC Waina i i sh Ne No More Changes Done date icon OEN ER arate Rois Mudd SIR Spite TENNER TEERIRANTA E l I Bi PFE vg i R k arsed Ri WEE ti gi if i4 t 3i 3 y ipit i Ay ie t ie f H real EG ipi A Ma Bis l itih 2 E 4 R i i jit j PANE pae T if ane Pate Status 3 bef ry tig Pea tad uh oe 4 ts ta Mb ae Oa alt YES BOP le PES of ii ES Status updates will display itt z f Wig hi bd p PARAL i E bind a Ma HS here ANANA R Strats k Da pis 1 Beetbibaee de ae itt ape aa On POR LELE Ha Pak tin bi E t De FA i ERE AAA EER T E A TEIR Bl ababat mts PRR IO Beha Esse ee TEI IEF EEE IT IFINI IERI TEIE S SEDI PEDERED Ch UIP LL ER DP EWORRES g LEE TIER TRE BER aed TIN at i N 16 The reassignment module offers a number of options Reassign to existing cluster Highlighted syllables will be moved from their current cluster to the cluster name selected from the dropdown menu To use Select a syllable s then press the Reassign button below VoICE will reassign the syllable s then re launch the reassignment module window Create a new cluster Highlighted syllables will be moved from their current cluster to a new cluster with a name from the dropdown m
10. Introduction The walkthoughs for VoICE are in the following pages It is strongly recommend to do them in order before proceeding with your own analyses In the first tutorial Clustering a Single Recording Session will present a walkthrough of how to cluster a single recording session The data for this first recording session are included in the VoICE download zip file and are in a directory entitled sample_data containing recordings from a bird Yellow119 The second tutorial Assigning New Recordings to Existing Clusters will contain a walkthrough on how to assign syllables from a second recording session to the clusters we create in the first tutorial For simplicity we will assign the first recording session to itself The screenshots here are from the Mac OS X version of VoICE but the Windows version should be nearly identical Please be sure that you have VoICE and its dependencies properly installed before beginning these tutorials Clustering a Single Recording Session The data you start with is a folder syllable containing a syllable table XLS file constructed in Sound Analysis Pro and WAV files from which the syllable table was constructed Each recording session must fit this same format Note A feature batch created by either the Feature Batch module or Explore and Score module is acceptable We have determined that while more time consuming a Feature Batch created by Explo
11. Vocal Inventory Clustering Engine VoICE Manual and Walkthrough TABLE OF CONTENTS Installation Instructions Windows BY 60 k Ge cnet eran eR ROO eRe ea eR er Tne eee ee ete ee eee eee tere 3 VOICE USV oreren N acieenned ued iain R E on ede nase nce 4 Mac OS X MOC cei cael a ccc art canaries E TEE E E aed mineecense tastes 5 VOICE USV ieii a E id cones tan ee sie cua ee waa heat ate adctn a uaaaie ete tals 6 Walkthroughs VoICE IITHOGUICH OM ais tekesetesecstivmateths E E waa aees even etn E T Clustering a Single Recording Session ccccccccecccececceceeeceeecuecsaeeceeceeeeeeesseeseesaeeaaeegeesaeseges 8 Step 1 Similarity Batch Module cccccccccccceecceeeceeeceeceeeceeeceeeceeesseesaeesaeesaeeceeeseeegeeegs 9 Step 2 Determine Merging Threshold cccccccccseccecceeeeseeeeeeeeeeseueeeeeeeeeseeeseeeseeeeeess 13 Step 6 REAaSSIGN SVIlADISS onera osc teetecietessauueteeGahemr E ties 16 Assign New Recordings to Existing Clusters 0 0 0 0 ccccc cece eeeeeeeeseeeseeeaeeeaeeeseeeseeeseeeseeeaeeeaeeeaes 19 step TE Score OITA othe betas ci sthe irate r AE E hele adie ehieheaneeee 21 Step 2 Select a Global Similarity GS Threshold For Assignment cc ecceeeeeeees 23 Step 3 Navigate Through Modules ccccccccceccceeceeeceeeeseeceueceuesueceueseueceeessueseeesaess 25 TAE TIGDFEAKING WOU Gt 4 osis se oiace eas E R A canes 25 The Novel Syllable Derivation Module cccc
12. arallel Processing Toolbox will increase the speed of the similarity batch by as many fold as there are usable processing cores The Run Similarity Batch button turns green when the batch is complete Depending on the number of syllables in the batch and the processing capability of your machine this can be a very time consuming process Note The software may appear to be unresponsive upon initially clicking Run Similarity Batch If no error message appears in red at the MATLAB Command Window be assured the software is running properly note the Busy notation in the bottom left hand corner of the command window 9 lt Student Version gt similarity_module Similarity Batch Module Select Syllable Tabie Current Syllable Table EA Similarity Scoring Clustering Cluster Syllables Min Dur 6 Determine Merging Threshold Win Size 41 Status Your similarity batch is running This may take a while A status bar will spawn in MATLAB Desktop Once the similarity batch has completed and its button has turned green hit Cluster Syllables The button will turn yellow and a separate status bar will launch as the clustering and dendrogram trimming steps occur oe lt Student Version gt similarity_module Similarity Batch Module Select Syllable Table Current Sylable Table MVolumesfitacintosn HOAJsors zburketvDosktoplsampio_data Yello
13. become active lt Student Version gt assignment_module VoICE Finch Assignment Module Syllables from the assignment directory will be compared to the existing clusters in the reference directory and given the same cluster designation should they pass a user defined global similarity GS threshold Assianment Directory Reference Directory Select A Feature Batch Select Directory Current Feature Batch Current Directory Similarity Batch m Assignment Options Enter a percentage of reference directory After completion of the similarity batch enter a clusters to score assignment directory global similarity threshold a syllable must syllables against A higher percentage attain with a reference cluster in order to be increases processing time but may assigned to that cluster provide more accurate assignment Assignment GS Threshold 50 Reference cluster 50 At this threshold you will have Min Dur 6 assigned to existing clusters Win Size 74777 to manually tiebreak passed into novel cluster detection Run Similarity Batch Assign Syllables Status Valid assignment feature batch and valid reference directory selected Similarity batch can now be run 21 Before running the similarity batch consider the following in the Similarity Batch panel Reference cluster This setting dictates what percentage of the syllables in eac
14. cccccecceeeeeeeeeeeeeeeceeeteeeseeeteeeaess 26 VoICE USV Halil g eyo Ue g emrrenperte tenes es ete ear Renter Rye E ee eee een 28 Step 1 Score Similarity Between USVS ccccccccecccseccceeceeeceeeceeeceeeceeeseeeceeseeesseesaeeseeeseeeseegs 30 Step 2 Assign Clustered Syllables to Canonical Call Type s cccccccecceeeceeeeeeeeeeeeeteeeneees 31 The Assignment Module cccccceccecceccecceeceeceeceeeseccueseeceeceeceeseeeueseusaeecercesseesenseeses 32 VoICE Windows Installation Install MATLAB R2014a with Signal Processing Toolbox and optionally Parallel Computing Toolbox Unzip VoICE zip to its own directory and add this directory to the MATLAB Path Install ImageMagick http www imagemagick org script binary releases php windows make sure to leave the add to system path option checked during install Rename the convert application in the install directory to imconvert Install R 3 1 2 http r project org and add the bin folder in the install directory to the system Path variable See hittps www java com en download help path xml for details specific to your Windows version regarding modification of the Windows path Install a Perl interpreter if you don t have one already recommend Strawberry Perl http strawberryperl com Add the bin folder in the install directory to the system Path variable using the same procedure as performed for R
15. d to that cluster provide more accurate assignment Assignment GS Threshold 50 Reference cluster 50 At this threshold you will have Min Dur 6 assigned to existing clusters to manually tiebreak passed into novel cluster detection Your similarity batch is running This may take a while A status Dar will spawn in MATLAB Desktop after a few moments Win Size 41 Status 6 gt 22 Step 2 Select a Global Similarity GS Threshold for Assignment When the similarity batch completes the Assignment Options panel becomes active The information in this panel determines the level of user involvement in finalizing cluster assignments A user editable field for Assignment GS Threshold which defaults to 50 can be altered This threshold is the level of average global similarity a syllable must reach with an existing cluster in order to be considered for assignment to it Consider the following when determining a GS threshold 1 A high GS threshold requires a very good match which will lead to assignments that the user can have great confidence in This will likely result in fewer automatic assignments 2 A low GS threshold will lead to many automatic assignments but the user will need to have lower confidence in these as less acoustic similarity is required for assignment 3 Regardless of GS threshold chosen the user will be able to in the final step of the w
16. e window All of the syllables for each syllable type as determined by the merging threshold selected in the previous window will be presented together in a spectrogram The user can use the zoom and drag tools see top left of window to navigate through the clusters Note it is highly advised to use horizontal zoom only which can be accomplished by first selecting the zoom in tool and right clicking on a spectrogram then navigating to zoom options and selecting horizontal zoom When the zoom or drag tools are not selected the user is able to click on individual syllables within each cluster which will highlight them in gray lt Student Version gt reassign_syllables om nh Ca F VoICE Finch Syllable Reassignment Module While VoICE is designed to cluster and assign syllables at a very low error rate occasional miscategorization occurs Use this module to correct those errors by clicking on misplaced syllables and using the dropdown menu to reassign them to the correct cluster See manual for further detail on more advanced features Tip Use horizontal only zoom when zooming on spectrograms Get Syll IDs eae m Reassign to existing cluster x107 saddlebrown Select a cluster sl m Create a new cluster Create a cluster gt o S g Reassign m Find Subtypes Select a cluster al 1 e a
17. e te tee ss 17 ats TEE LEN 110 oat Hee a o a o e CE Hoe hee tbe bi eee 10 1400 1h 140 14K S nee es i i i Of ae 2 N R F pA iu d A 120 fa as t E IC 7 E3 09 939 Lom Leda Lind 116r Low Wet Tare TEC leew LIES eco EG 169 1653 2705 1730 13m 1835 Lat 13r es lias les Dreetad Ey Sok lesda hi ir par tay ila sasali da nan Ilaj lle z 2 2 2 m e i2 a au oe lt 2 on i a i Pe 5 m 51 m a 5 s z co 8 no 3 2 no t 3 m g seH 2 2 A ka a m 3 4 P s pr co 32 2 J x g i hi J F mEn 3 33 32 mei l iin iv in lt 115 iin Er iin l5 be U a 10 1x 2 10 te 1 are Peer Ii ea s eis Foal et are trs irs irs irs iwe irs areal ly a Arabaj Ly a Li yaler sy tU Lrsalew sy 52 Aroa ly ae Li aaler sy t Status m Syllable belonas in cluster Tiebreaking syllable 1 of 6 novel a Confirm Back Finalize Use of this module is mostly self explanatory Of note the dropdown menu displays the average GS between a given syllable and each cluster The user may also deem syllables as novel in which case they will be passed through to novel syllable type detection See next Syllable belonas in cluster CACO Coni black 17 36 blue 9 8 cyan 42 87 darkmagenta 29 79 J floralwhite 50 53 saddlebrown 26 05 Once all assignments are complete the Finalize button becomes active and the user will proceed on t
18. enu of the user s choice To use Select a syllable s then select a cluster name from the dropdown menu within the Reassign to existing cluster pane Press the Reassign button below VoICE will move the selected syllable s to a new cluster then re launch the reassignment module window Find subtypes The cluster selected in this dropdown menu will be split into a user dictated number of individual clusters based on similarity relationships within the cluster To use within the Find Subtypes panel select a cluster from the Select a cluster dropdown Next choose a number of subtypes to be returned from the Choose a number dropdown that has been activated Finally hit Go VoICE will probe for the selected number of subtypes then re launch the reassignment module window with the selected cluster divided into the number of requested subtypes Get syllable IDs Not currently functional Delete syllables Highlighted syllables will be deleted from your dataset Your original data will be preserved in separate Matlab R files in the event you make an error No More Changes Done VoICE will perform final calculations and close the reassignment module In our example no reassignments are necessary We encourage the user to explore the various options using different merging thresholds on their own in order to gain familiarity with how the software works When the reassignment module is closed the analy
19. er or lower Proceed with assigning calls to their canonical categories as defined by Scattoni et al in 2008 Note detailed descriptions of the call categories are in this manuscript When all syllables are assigned the buttons will turn gray and new options will appear Frequency Complex Harmonic Two Syllable lt Student Version gt voice_usv_assign 0 01 0 015 0 02 0 025 0 03 0 035 0 04 0 045 Time Working On File 36 of 36 Chevron Short Freq Step Downward Double Misc 32 First hit the red Finalize Assignments button New options will then appear 9 lt Student Version gt voice_usv_assign Frequency 0 005 0 01 0 015 0 02 0 025 0 03 0 035 0 04 0 045 Time Working On File 36 Of 36 Back Create Cluster WAV Files will generate four folders within the directory for each animal in the analysis 1 joined_clusters_clusters A directory containing a single WAV file containing all of the calls for each cluster as determined by the automated tree trimming algorithm 2 sorted_syllables_clusters A directory containing subdirectories for each cluster as determined by the automated tree trimming algorithm with a WAV file for each call in that cluster 3 joined_clusters_pie A directory containing a single WAV file containing all of the calls for each call type as determined by the user assignments in the assignment module 4 sorted_syllables_pie A directory containing subd
20. f the table presented above Iterative Tree Trimming Curve 40 30 Cluster n 1 merging threshold Use the dropdown menu to select a merging threshold for this tutorial select 0 31 Press Generate Clusters The button will turn yellow while processing then green when complete lt Student Version gt determine_merging_ threshold Threshold black IGS black n blue IGS blue n cyan IGS cyann darkmagenta IGS darkmagenta n floralwhite IGS floralwhite n lightcyan IGS lig 3 0 1700 88 7668 47 87 9511 47 86 6048 87 7429 aS 72 7346 60 88 8735 4 0 3100 88 7668 47 87 9511 47 86 6048 13 87 7429 47 64 6143 107 76 9821 5 0 4500 88 7668 47 87 9511 47 86 6048 13 87 7429 47 56 3791 178 0 iterative Tree Trimming Curve 0 31 a Q peen m Reassign Syllables amp Cr oO i Status Clusters generated Click Reassign Syllables to view edit BEWARE Hitting Generate Clusters again will wipe all reassignments 0 0 0 2 0 4 0 6 0 8 1 0 1 merging threshold 15 Step 3 Reassign Syllables Once clusters have been generated at the desired merging threshold proceed to the syllable reassignment module by clicking Reassign Syllables The current window will not close automatically which will allow you to apply a new merging threshold should the clusters appear undesirable in the reassignment module The syllable reassignment module will display all the clusters in a scrollabl
21. h cluster from the first recording will be used for similarity scoring These clusters should be highly homogeneous thereby making comparison with every single syllable in each cluster somewhat redundant A higher percentage will certainly not yield poorer results but will certainly increase processing time We set the default to 50 and will use it for the tutorial The Run Similarity Batch button will turn yellow and a progress bar will appear in the MATLAB Command Window while the similarity batch runs This may take some time The progress bar will reach 100 and the button will turn green when the similarity batch is complete 9 lt Student Version gt assignment_module VoICE Finch Assignment Module Syllables from the assignment directory will be compared to the existing clusters in the reference directory and given the same cluster designation should they pass a user defined global similarity GS threshold Assianment Directory Reference Directory Select A Feature Batch Select Directory Current Feature Batch Current Directory Similarity Batch Assianment Options Enter a percentage of reference directory After completion of the similarity batch enter a clusters to score assignment directory global similarity threshold a syllable must syllables against A higher percentage attain with a reference cluster in order to be increases processing time but may assigne
22. irectories for each call type as determined by the user assignments in the assignment module with a WAV file for each call of that call type Create Pie Charts will generate pieChart pdf within the directory for each animal in the analysis This chart displays the percentage distribution of each call type as determined by the user assignments in the assignment module This concludes the VoICE USV tutorial 33
23. mpute and preprocessCore packages source http www bioconductor org biocLite R biocLite impute biocLite preprocessCore Install SoX http sox sourceforge net then add the install directory to the system Path variable see Path modification instructions for R installation above Launch MATLAB then type voice _usv at the command line to launch the GUI VoICE Mac OS X Installation Install MATLAB R2014a with Signal Processing Toolbox and optionally Parallel Computing Toolbox Unzip VoICE zip to its own directory and add this directory to the MATLAB Path Install R 3 1 2 http r project org Launch R then 1 Install the GO db package source http www bioconductor org biocLite R biocLite GO db 2 Install the WGCNA package install packages WGCNA 3 Install the gdata package install packages gdata 4 Install the impute and preprocessCore packages from bioconductor source http www bioconductor org biocLite R biocLite impute biocLite preprocessCore 5 Install the ggmap and png packages install packages ggmap install packages png Install Homebrew Note Homebrew is a free command line package manager for OSX It is by no means the only way to install the software that reference below It is the easiest way to install this software and have it placed where it needs to be for VoICE to run properly Therefo
24. ng threshold At the top a field is displayed describing the merging thresholds at which cluster N was stable over at least one merge after that threshold The first column in this field Threshold is the Pearson correlation subtracted from 1 at which stability was first achieved Subsequent columns should be viewed in pairs Each pair of columns describe one cluster where cluster names are unique colors The IGS is the intracluster global similarity a measure of how homogeneous the cluster is on a O to 100 scale The n is the number of syllables in the cluster Threshold black IGS black n blue IGS blue n cyan IGS cyann darkmagenta IGS darkmagenta n floralwhite IGS floralwhite n lightcyan IGS lic 0 0900 88 7668 47 87 9511 47 86 6048 13 87 7429 47 80 2325 13 82 4429 0 1100 88 7668 47 87 9511 47 86 6048 13 87 7429 47 80 2325 13 88 8735 0 1700 88 7668 47 87 9511 47 86 6048 13 87 7429 47 72 7346 60 88 8735 13 Below in the same window an image is displayed plotting the relationship between 1 merging threshold and the number of clusters generated at this threshold Points where the curve is flat represent points at which the cluster N remained stable over at least two merging thresholds These are the points at which increased tolerance for variability in merging clusters was allowed yet no clusters merged together indicating potentially stable configurations of the animal s repertoire The flat points in the curve correspond to the rows o
25. o the next module 25 The Novel Syllable Derivation Module If novel syllables are deemed to be present in the Assignment Module and or the Tiebreaking Module this module will launch Otherwise it will be skipped and the reassignment module will launch in its place A blank module will open then the user must click the Derive Novel Syllables button A similarity batch between the novel syllables will run and the user will be prompted to select a merging threshold in a process similar to the one described for Determine Merging Threshold See page 13 for more information 9 lt Student Version gt novelty_module Threshold grey IGS greyn turquoise IGS turquoise n 1 0 0100 0 1 49 1753 3 VoICE Finch Novel Syllable Derivation Module Press the Derive Novel Syllables button at left select a merging threshold then generate clusters at the desired merging threshold Use the reassign syllables module to inspect the novel clusters Once the ideal threshold has been selected proceed to reassignment and close this window Iterative Tree Trimming Curve Select a Meraina Threshold Select a threshold 2 5 Generate Clusters Reassign Syllables 2 0 Cluster n Status Similarity batch between novel Te syllables is complete Choose a a merging threshold a rr a ss as Pi 0 0 0 2 0 4 0 6 0 8 1 0 1 merging threshold In the case of our example since only four syllables were put
26. orkflow reassign syllables that are placed incorrectly The assignment module will display the number of automatic assignments manual tiebreaks and syllables that are considered novel at whichever GS threshold the user selects 9 lt Student Version gt assignment_module VoICE Finch Assignment Module m Assianment Directory Select A Feature Batch Current Feature Batch Similarity Batch Enter a percentage of reference directory clusters to score assignment directory syllables against A higher percentage increases processing time but may provide more accurate assignment Reference cluster 50 Min Dur 6 Win Size 41 Syllables from the assignment directory will be compared to the existing clusters in the reference directory and given the same cluster designation should they pass a user defined global similarity GS threshold r Reference Directory Select Directory Current Directory Assignment Options After completion of the similarity batch enter a global similarity threshold a syllable must attain with a reference cluster in order to be assigned to that cluster Assignment GS Threshold 50 At this threshold you will have 322 assigned to existing clusters 6 to manually tiebreak 4 passed into novel cluster detection Assign Syllables Status Now showing assignment breakdown for GS threshoid 50 Ready to assign
27. re will reference only Homebrew for this installation guide Troubleshooting Homebrew installation or other methods of installing the proceeding software is beyond what can support as an author If you should run into trouble and searching the Internet for answers is not fruitful can try to help via email 1 Install Homebrew by opening Terminal and pasting the below at the command prompt ruby e curl fsSL https raw githubusercontent com Homebrew install master install 2 Install SoX by typing the following at the command prompt brew install sox 3 Install ImageMagick by typing the following at the command prompt brew install imagemagick Launch MATLAB then type voice at the command line to launch the GUI VoICE USV Mac OS X Installation Install MATLAB R2014a with Signal Processing Toolbox Unzip VoICE_usv zip to its own directory and add this directory to the MATLAB Path Install R 3 1 2 http r project org Launch R then 1 Install the GO db package source http www bioconductor org biocLite R biocLite GO db 2 Install the WGCNA package install packages WGCNA 3 Install the impute and preprocessCore packages source http www bioconductor org biocLite R biocLite impute biocLite preprocessCore Use Homebrew see above to install Sox Launch MATLAB then type voice _usv at the command line to launch the GUI VoICE Walkthrough
28. re and Score results in a cleaner dataset D sample_data s Em io Be Shared Folder 3 v Name Date Modified Size Kind ir Yellow119_122013 xls Dec 11 2014 11 29 AM 114 KB Micros ksheet Yellow119_Decemb 013_29664384 wav Dec 20 2013 8 14 AM 1 6 MB Waveform audio Yellow119 Decemb 013_29684562 wav Dec 20 2013 8 14 AM 413 KB Waveform audio Yellow119_Decemb 013_29691843 wav Dec 20 2013 8 15 AM 1 3 MB Waveform audio Yellow119_Decemb 013_29705993 wav Dec 20 2013 8 15 AM 1 8 MB Waveform audio Yellow119_ Decemb 013_29745682 wav Dec 20 2013 8 16 AM 1 8 MB Waveform audio Yellow119_Decemb 013_29765924 wav Dec 20 2013 8 16 AM 1 2 MB Waveform audio Yellow119_Decemb 013_29779153 wav Dec 20 2013 8 16 AM 1 8 MB Waveform audio Yellow119 Decemb 013_29799624 wav Dec 20 2013 8 16 AM 1 5 MB Waveform audio Yellow119_Decemb 013_29818694 wav Dec 20 2013 8 17 AM 552 KB Waveform audio Yellow119_ Decemb 013_29828574 wav Dec 20 2013 8 17 AM 625 KB Waveform audio Yellow119_ Decemb 013_29836554 wav Dec 20 2013 8 17 AM 841 KB Waveform audio Launch VoICE by typing voice at the MATLAB command line Click Cluster a Single Recording Session eo lt Student Version gt voice VoICE Finch Cluster a Single Recording Session Compare Two Recording Sessions Analysis Functions Step 1 Similarity Batch Module The Similarity Batch Module will launch Use the
29. s We have used custom written MATLAB code to generate these wav files that we do not provide in the software package M sample_data_usv ss Egon ior By HY Q Shared Folder Name Size Kind Ol wav 6 KB Waveform audio 02 wav 28 KB Waveform audio O03 wav 6 KB Waveform audio 04 wav 5 KB Waveform audio 05 wav 8 KB Waveform audio O06 wav 12 KB Waveform audio gt O7 wav 7 KB Waveform audio O8 wav 4KB Waveform audio O09 wav 23 KB Waveform audio 10 wav 23 KB Waveform audio 11 wav 11 KB Waveform audio gt 12 wav 2 KB Waveform audio 13 wav 13 KB Waveform audio gt 14 wav 3 KB Waveform audio 15 wav 9 KB Waveform audio 16 wav 24 KB Waveform audio 17 wav 24 KB Waveform audio 18 wav 34 KB Waveform audio 19 wav 34 KB Waveform audio 20 wav 33 KB Waveform audio Step 1 Score Similarity Between USVs VoiCE USV is contained within two modules To launch type voice_usv at the MATLAB command prompt The interface will then open o 9 lt Student Version gt voice_usv VoICE USV m Similarity Scorin Assiaqnment Animal 1 meone Enter Animal ID Animal ID Assianment Animal 2 Sexe int Assianment Options Select a Similarity Batch no NY Select a Similarity Batch Select a folder ree Cohesion Threshold 0 80 Min Cluster Size 5 Launch Assignment Module Press Select a folder and then navigate to sample _data_usv and hit Open
30. sis is considered complete Close the Determine Merging Threshold window When finished your original data folder will contain new items Many are internal to VoICE s function The ones containing data relevant to the result of clustering are summarized here syntax_summary csv A comma separated file containing a transition probability table and syntax entropy and stereotypy scores for the dataset that was clustered cluster_tables A folder containing csv files named by cluster with the acoustic data for each syllable in the cluster joined_clusters A folder containing joined wav files of all the syllables in each cluster sorted_syllables A folder containing subfolders named by cluster containing the individual wav files for all of the syllables in each cluster cluster_dendrogram pdf A dendrogram where leaves represent syllables and color stripes below corresponding to the cluster assignments for each syllable Note It is strongly suggested that the user not remove or rename files from this directory Instead copy files to new locations for further analysis This concludes the Clustering a Single Recording Session tutorial Assigning New Recordings to Existing Clusters Here we will assign the syllables from a second recording session to the clusters created in the previous tutorial As an example this situation would arise if the user were to record and cluster the vocalizations from a bird one day
31. subfolders named by cluster containing the individual wav files for all of the syllables in each cluster Note It is strongly suggested that the user not remove or rename files from this directory Instead copy files to new locations for further analysis This concludes the Assigning New Recordings to Existing Clusters tutorial 27 VoICE USV Walkthrough Introduction The walkthough for VoICE USV are in the following pages strongly recommend your doing the tutorial before proceeding with your own data In this tutorial will score similarity between the vocalizations of one mouse and then assign them to canonical call types VoICE USV offers the option to blind the user to the animal s genotype which will not do here The sample data are included in the VoICE USV download zip file in a folder entitled sample _data_usv The screenshots here are from the Mac OS X version of VolCE USV but the Windows version should be nearly identical Please be sure that you have VolCE USV and its dependencies properly installed before beginning these tutorials 28 VoICE USV Walkthrough The data you start with is a folder containing USVs in individual WAV files This folder is entitled sample _data_usv It contains 20 USVs each in their own WAV files Note When collecting your USVs please ensure that filenames are sequential We do not have a recommendation as how to generate your individual WAV file
32. the existing clusters in the reference directory and given the same cluster designation should they pass a user defined global similarity GS threshold Assianment Directory Reference Directory Select A Feature Batch Similarity Bateh 4 Enter a percentage of reference directory After completion of the similarity batch enter a clusters to score assignment directory global similarity threshold a syllable must syllables against A higher percentage attain with a reference cluster in order to be increases processing time but may assigned to that cluster provide more accurate assignment Assignment GS Threshold 50 Reference cluster 50 At this threshold you will have Min Dur assigned to existing clusters Win Size to manually tiebreak passed into novel cluster detection Run Similarity Batch Assign Syllables Status SSMS Status Updates Will Display Here 20 Step 1 Score Similarity The Assignment Directory contains the to be assigned syllables this is sample_data 2 The Reference Directory contains the already clustered syllables this is sample _data Press Select A Feature Batch in the Assignment Directory panel then navigate to the feature batch XLS file in sample_data 2 Press Select Directory in the Reference Directory panel then navigate to the sample_data directory The Run Similarity Batch button will
33. w119 122019318 Similarity Scoring Clustering Determine Merging Threshold Syllables being clustered a status bar will spawn When the Cluster Syllables button turns green you can proceed to determining the merging threshold which will then yield the syllable clusters by clicking Determine Merging Threshold 09 lt Student Version gt similarity_module Similarity Batch Module Select Syllable Table Current Syllable Tabie E PAA Similarity Scoring Clustering Determine Merging Threshok Clustering and iterative trimming complete Step 2 Determine Merging Threshold After clicking Determine Merging Threshold a new window opens which contains information regarding the clusters at a number of merging thresholds ee lt Student Version gt determine_merging_ threshold Threshold black IGS black n blue IGS blue n cyan IGS cyann darkmagenta IGS darkmagenta n floralwhite IGS floralwhite n lightcyan IGS lig 1 0 0900 88 7668 47 87 9511 47 86 6048 13 87 7429 47 80 2325 13 82 4429 2 0 1100 88 7668 47 87 9511 47 86 6048 13 87 7429 47 80 2325 13 88 8735 3 0 1700 88 7668 47 87 9511 47 86 6048 13 87 7429 47 72 7346 60 88 8735 Iterative Tree Trimming Curve Select a threshold 2 wv Generate Clusters e oOo m g e o A a e oO p ry Status epteh SASEREEEE SES Status Updates Here a SS trevercrers i as ae 0 0 0 2 0 4 0 6 0 8 1 0 1 mergi

VoICE Manual

Contents

Download Pdf Manuals

Related Search

Related Contents