PD_RewP_Script.m

%%

% N=28 Parkinson's patients and N=28 matched controls
% PD patients came in for 2 sessions 1 week apart: ON or OFF meds (counterbalanced).  
% EEG files are labeled with session #, see ONOFF.mat for which session was ON or OFF.  
% Note that controls DID NOT have 2 sessions.


clear all; clc
homedir='Y:\EEG_Data\PDDys\PD 4 PREDICT\';
datalocation=[homedir,'\PROCESSED EEG DATA\'];   % Data are here
savepath = [datalocation,'CLEAN\'];
cd(datalocation);

load([homedir,'VV_Behavior.mat']); % Aggregate behavior files output from Matlab Psychtoolbox
load([homedir,'ONOFF.mat']);  % 3 columns: subject, session, [ON=1 OFF=0]
load([homedir,'BV_Chanlocs_60.mat']);
%*************************************

MEASURES = xlsread([homedir,'MEASURES']);  % Subj symptom measures taken in ON session
%************************
%   COLUMN LABELS
%   MEASURES(:,1) = PD IDx
%   MEASURES(:,2) = NAART Scores
%   MEASURES(:,3) = BDI Ratings
%   MEASURES(:,4) = MMSE Scores
%   MEASURES(:,5) = UPDRS Ratings
%   MEASURES(:,6) = Years Since Diagnosis (Rank Ordered)
%   MEASURES(:,7) = Levadopa Equivalent Dose (LED)
%   MEASURES(:,8) = Accelerometer hand placement: 1 = Left Hand // 2 = Right hand
%************************

% Subject Numbers
PDsx=[801:811,813:829];  
CTLsx=[8010,8070,8060,890:914];


%% MAKE ERPs
for subj=[PDsx,CTLsx]
    for session=1:2;
        if (subj>850 && session==1) || subj<850  % If not CTL, do session 2  (CTL did not have a session 2)
                
                load([num2str(subj),'_Session_',num2str(session),'_PDDys_VV_withcueinfo.mat'],'EEG','bad_chans','bad_epochs','bad_ICAs');
                
                for ai=1:size(EEG.epoch,2)
                    VECTOR(ai,1)=EEG.epoch(ai).FB;
                    VECTOR(ai,2)=EEG.epoch(ai).Resp;
                    VECTOR(ai,3)=EEG.epoch(ai).Resptime;
                    VECTOR(ai,4)=EEG.epoch(ai).Stim;
                    VECTOR(ai,5)=EEG.epoch(ai).Stimtime;
                    VECTOR(ai,6)=EEG.epoch(ai).Cie;
                    VECTOR(ai,7)=EEG.epoch(ai).Cuetime;
                    VECTOR(ai,8)=EEG.epoch(ai).RT;
                    VECTOR(ai,9)=EEG.epoch(ai).BEHCondi;
                    VECTOR(ai,10)=EEG.epoch(ai).BEHOptimal;
                    VECTOR(ai,11)=EEG.epoch(ai).BEHRT;
                    VECTOR(ai,12)=EEG.epoch(ai).BEHFB;
                end
                % Remove practice trials
                VECTOR(isnan(VECTOR(:,9)),:)=NaN;
                % Add this for later: FB-parsed by condi
                for vvi=1:length(VECTOR),
                    if VECTOR(vvi,1)==0, VECTOR(vvi,13)=VECTOR(vvi,9);
                    elseif VECTOR(vvi,1)==1, VECTOR(vvi,13)=4+VECTOR(vvi,9);
                    end
                end
                
                % Remove the bad ICAs identified by APPLE:
                bad_ICAs_To_Remove=bad_ICAs{2};
                EEG = pop_subcomp( EEG, bad_ICAs_To_Remove, 0);
                
                % low-pass filter for display
                dims=size(EEG.data);
                EEG.data=eegfilt(EEG.data,500,[],20);
                EEG.data=reshape(EEG.data,dims(1),dims(2),dims(3));
                
                % Set times
                tx=-6000:2:1998;
                b1=find(tx==-200);  b2=find(tx==0);
                t1=find(tx==-500);  t2=find(tx==1000); % For ERPs
                r1=find(tx==250);  r2=find(tx==450); % For Topos
                tx2disp=-500:2:1000;
                
                % Accelerometer was worn on the non-dominant hand
                % Aggregate accelerometer data
                EEG.X=EEG.X-repmat(mean(EEG.X),4000,1);
                EEG.Y=EEG.Y-repmat(mean(EEG.Y),4000,1);
                EEG.Z=EEG.Z-repmat(mean(EEG.Z),4000,1);
                % Add to EEG.data as 61st channel - but not the rejected trials
                EEG.data(61,:,:)=(EEG.X(:,bad_epochs{1}~=1).^2)+(EEG.Y(:,bad_epochs{1}~=1).^2)+(EEG.Z(:,bad_epochs{1}~=1).^2);
                dims=size(EEG.data);

                % Basecor your ERPs here so they are pretty.
                BASE1=squeeze(  mean(EEG.data(:,b1:b2,:),2)  );
                for chani=1:dims(1)-1   % don't basecor the tremor data
                    DATA(chani,:,:)=squeeze(EEG.data(chani,:,:))-repmat( BASE1(chani,:),dims(2),1 );
                end
                
                % Parse by condition
                    %************************
                    %   CONDITIONS 
                    %       1 = CHOOSE  EASY  LOSE
                    %       2 = CHOOSE  HARD  LOSE
                    %       3 = MATCH   EASY  LOSE
                    %       4 = MATCH   HARD  LOSE
                    %       5 = CHOOSE  EASY  WIN
                    %       6 = CHOOSE  HARD  WIN
                    %       7 = MATCH   EASY  WIN
                    %       8 = MATCH   HARD  WIN   
                    %************************                
                for ai=1:8 % all FB
                    ERP(:,ai,:)=mean(DATA(:,t1:t2,VECTOR(:,13)==ai),3); % DATA(ELECTRODE, TIME , CONDITION)
                    TOPO(:,ai) = squeeze(mean(mean(DATA(:,r1:r2,VECTOR(:,13)==ai),2),3));                    
                end

                % Save and move on to next
                save([savepath,num2str(subj),'_Session_',num2str(session),'_PDDys_VV_withcueinfo_ALL_THE_GOODS.mat'],'ERP','TOPO','VECTOR');
                
                clc;
                disp(['AND PARTICPANT ',num2str(subj),' HAS BEEN SAVED']);
                
                clearvars -except datalocation ONOFF VV_Behavior BV_Chanlocs_60 PDsx CTLsx session subj savepath;
                close all;
                
        end
    end
end

%% COMBINE ERPS

site = [21,36];%  Cz FCz 
tx=-500:2:1000;
time1 = 250;    time2 = 450;
t1=find(tx==time1);  t2=find(tx==time2); % FOR REW-P
TIME = time2-time1;
tx2disp=-500:2:1000;
COLS={'r','b','g','k'};

BigN=size(ONOFF,1)./2;

row=1;
for mi=1:size(ONOFF,1)
    disp([num2str(ONOFF(mi,1)),'_Session_',num2str(ONOFF(mi,2)),'_PDDys_VV_withcueinfo_ALL_THE_GOODS.mat']);
    load([savepath,num2str(ONOFF(mi,1)),'_Session_',num2str(ONOFF(mi,2)),'_PDDys_VV_withcueinfo_ALL_THE_GOODS.mat']);
    
    if ONOFF(mi,3)==1 % ON
        ON.ID(floor(row))=ONOFF(mi,1);
        ON.Session(floor(row))=ONOFF(mi,2);
        ON.VECTOR=VECTOR;
        ON.ERPs(floor(row),:,:)=squeeze(mean(ERP(site,:,:),1));
        ON.Topos(floor(row),:,:)=TOPO(:,:);
    elseif ONOFF(mi,3)==0 % OFF
        OFF.ID(floor(row))=ONOFF(mi,1);
        OFF.Session(floor(row))=ONOFF(mi,2);
        OFF.VECTOR=VECTOR;
        OFF.ERPs(floor(row),:,:)=squeeze(mean(ERP(site,:,:),1));
        OFF.Topos(floor(row),:,:)=TOPO(:,:);
    end
    row=row+.5;
    clear ERPs VECTOR;
end

row=1;
for CTLi=[8010,8060,8070,890:914];
    disp([num2str(CTLi),'_Session_1_PDDys_VV_withcueinfo_ALL_THE_GOODS.mat']);
    load([savepath,num2str(CTLi),'_Session_1_PDDys_VV_withcueinfo_ALL_THE_GOODS.mat']);
    CTL.ID(floor(row))=CTLi;
    CTL.Session(floor(row))=1;
    CTL.VECTOR=VECTOR;
    CTL.ERPs(floor(row),:,:)=squeeze(mean(ERP(site,:,:),1)); 
    CTL.Topos(floor(row),:,:)=TOPO(:,:);
    
    row=row+1;
    clear ERPs VECTOR;
end
CtlN=row-1;

%************************
%         TOPOS 
%************************
TOPO_CON = (CTL.Topos(:,:,5)+CTL.Topos(:,:,6)+CTL.Topos(:,:,7)+CTL.Topos(:,:,8)   )/4;
TOPO_ON = (ON.Topos(:,:,5)+ON.Topos(:,:,6)+ON.Topos(:,:,7)+ON.Topos(:,:,8)   )/4;
TOPO_OFF = (OFF.Topos(:,:,5)+OFF.Topos(:,:,6)+OFF.Topos(:,:,7)+OFF.Topos(:,:,8)   )/4;

figure; hold on

% CONTROL TOPO
subplot(1,3,1)
topoplot(mean(TOPO_CON,1),BV_Chanlocs_60);
title('CONTROL');
set(gca,'clim',[-3 3]);

% ON TOPO
subplot(1,3,2)
topoplot(mean(TOPO_ON,1),BV_Chanlocs_60);
title('ON');
set(gca,'clim',[-3 3]);

% OFF TOPO
subplot(1,3,3)
topoplot(mean(TOPO_OFF,1),BV_Chanlocs_60);
title('OFF');
set(gca,'clim',[-3 3]);
cbar

      
%************************
%   ERP 
%************************
win_CON = (CTL.ERPs(:,5,:)+CTL.ERPs(:,6,:)+CTL.ERPs(:,7,:)+CTL.ERPs(:,8,:))/4;
win_ON = (ON.ERPs(:,5,:)+ON.ERPs(:,6,:)+ON.ERPs(:,7,:)+ON.ERPs(:,8,:))/4;
win_OFF = (OFF.ERPs(:,5,:)+OFF.ERPs(:,6,:)+OFF.ERPs(:,7,:)+OFF.ERPs(:,8,:))/4; 

lose_CON = (CTL.ERPs(:,:,1)+CTL.ERPs(:,:,2)+CTL.ERPs(:,:,3)+CTL.ERPs(:,:,4))/4;
lose_ON = (ON.ERPs(:,:,1)+ON.ERPs(:,:,2)+ON.ERPs(:,:,3)+ON.ERPs(:,:,4))/4;
lose_OFF = (OFF.ERPs(:,:,1)+OFF.ERPs(:,:,2)+OFF.ERPs(:,:,3)+OFF.ERPs(:,:,4))/4;

figure;hold on;
rectangle('Position',[time1,0,TIME,3],'Curvature',0.1,'FaceColor',[.9 .9 .9])% ON WIN
plot(tx2disp,squeeze(nanmean(win_CON,1)),COLS{1});
plot(tx2disp,squeeze(nanmean(win_ON,1)),COLS{2});
plot(tx2disp,squeeze(nanmean(win_OFF,1)),COLS{3});
shadedErrorBar(tx2disp, squeeze(nanmean(win_CON,1)),  nanstd(squeeze(win_CON))   ./sqrt(28),COLS{1})
shadedErrorBar(tx2disp, squeeze(nanmean(win_ON,1)),  nanstd(squeeze(win_ON))   ./sqrt(28),COLS{2})
shadedErrorBar(tx2disp, squeeze(nanmean(win_OFF,1)),  nanstd(squeeze(win_OFF))   ./sqrt(28),COLS{3})
plot(tx2disp,squeeze(nanmean(win_CON,1)),COLS{1},'LineWidth',4);
plot(tx2disp,squeeze(nanmean(win_ON,1)),COLS{2},'LineWidth',4);
plot(tx2disp,squeeze(nanmean(win_OFF,1)),COLS{3},'LineWidth',4);

title('ERPs FOR WINS');
h_legend=legend({'HC','ON','OFF'});
set(h_legend,'FontSize',12);
plot([0 0],[-6 6],'k:');
set(gca,'ylim',[-1 4],'xlim',[-100 1000])
pcrit=.05;
[H,P,CI,STATS]=ttest(win_CON,win_ON);
P(P>pcrit)=NaN;  P(P<=pcrit)=1;
plot(tx2disp,-.5*squeeze(P),'k','linewidth',3);  clear H P CI STATS;
[H,P,CI,STATS]=ttest(win_CON,win_OFF);
P(P>pcrit)=NaN;  P(P<=pcrit)=1;
plot(tx2disp,-.7*squeeze(P),'r','linewidth',3);  clear H P CI STATS;

CONTROL_ERP = squeeze(mean(win_CON(:,:,t1:t2),3));
ON_ERP = squeeze(mean(win_ON(:,:,t1:t2),3));
OFF_ERP = squeeze(mean(win_OFF(:,:,t1:t2),3));

[H,P,CI,STATS]=ttest(CONTROL_ERP,ON_ERP)
text(.7,3.5,['CONTROL v. ON  t= ',num2str(STATS.tstat),'   p= ',num2str(P)])
[H,P,CI,STATS]=ttest(CONTROL_ERP,OFF_ERP)
text(.7,3.3,['CONTROL v. OFF  t= ',num2str(STATS.tstat),'   p= ',num2str(P)])
[H,P,CI,STATS]=ttest(ON_ERP,OFF_ERP)
text(.7,3.1,['ON v. OFF  t= ',num2str(STATS.tstat),'   p= ',num2str(P)])

%************************
%   FOR ANALYSIS 
%************************
% Run these in SPSS to determine that there's no interaction between group
% and volition / difficulty
for condi=1:8   
    SPSS_CONT(:,condi)= squeeze(nanmean(CTL.ERPs(:,condi,t1:t2),3));
    SPSS_ON(:,condi)= squeeze(nanmean(ON.ERPs(:,condi,t1:t2),3));
    SPSS_OFF(:,condi)= squeeze(nanmean(OFF.ERPs(:,condi,t1:t2),3));
end

% So then combine all rewards across volition and difficulty conditions
REWP_ON = [SPSS_ON(:,5),SPSS_ON(:,6),SPSS_ON(:,7),SPSS_ON(:,8)];
REWP_OFF = [SPSS_OFF(:,5),SPSS_OFF(:,6),SPSS_OFF(:,7),SPSS_OFF(:,8)];

ALL_REWP_ON = mean(REWP_ON,2);
ALL_REWP_OFF = mean(REWP_OFF,2);

%************************
%   CORRELATIONS 
%************************       
YrsDx=tiedrank(MEASURES(:,6),1);

figure;
hold on;
subplot(2,1,1)
scatter(YrsDx,ALL_REWP_ON,'MarkerEdgeColor',[0 .5 .5],...
              'MarkerFaceColor',[0 .7 .7],...
              'LineWidth',1.5)
lsline
title('ON: REW-P v. YRS DIAGNOSED ')
set(gca,'ylim',[-2 5])
[RHO,PVAL] = corr(ALL_REWP_ON,MEASURES(:,6),'TYPE','Spearman');
text(.7,4,['r= ',num2str(RHO),'   p= ',num2str(PVAL)])

subplot(2,1,2)
hold on;
scatter(YrsDx,ALL_REWP_OFF,'MarkerEdgeColor',[0 .5 .5],...
              'MarkerFaceColor',[0 .7 .7],...
              'LineWidth',1.5)
lsline
title('OFF: REW-P v. YRS DIAGNOSED ')
set(gca,'ylim',[-2 5])
[RHO,PVAL] = corr(ALL_REWP_OFF,MEASURES(:,6),'TYPE','Spearman');
text(.7,4,['r= ',num2str(RHO),'   p= ',num2str(PVAL)])

%%