ࡱ> `!hiALвO<@/xcdd``cb``baV d,FYzP1n:&B@?b O@=P5< %! vf: KX+ssr-^QXAky[- P.P56$b\=ڍ25?%45au`#kA| 8ߊ kQeCU sB6 ,ƐXg\4<^F&&\[:@ĞB 8@= EHX9и oY>x] |Tչ?>0ELp’DH`e B-"bP0`O\i^\ZmK+ϭ'\pb+hhh̜d͝Pɽ{9;۷sBHa?7y4HW3{ L<\B3ܭ$0> !Ey9M@!JB <;Se a8^?xN΋k|hJq4!^Cvi9xoP7C ~/Da~śaο.w#/? גzԑf(T8q:ik1Hib&=rq+DT4;|/L]XL6}x~x+w/yxFF~KXcA^fi;y:JjϱV(oj-+SM/Ko!0%&Oy9Oˀf#d),|B&@n"g !HjPŏE_xo 6Kw,bnXLk`ò,s:r%#JQrTPR@xi <_qVODxP)Ѹ >F@G4P@KL~8y"\zF؏==-x}M*7d i.B&ؘ6Zf|l9y^\I[j4X*=iЎA [rv/ l U6b0\y奻R%ºS,Uq* ?STRï$ 5jăcHK8% g!|`99?ֲvŒ:BR_v1"U7 #SZ#ӛHHpd Ael\)l~Lƕ\|O%,C27бq(X啕:!*AS硕A΁ 0.WEU_n18c.Qb[ć\ٮo km xc;||Ow~O1Ga>sxC<-r ?.Qsub]sd]komO{/_c}|Ν:Fw'j4b`nN%kieeکuVy͂u?& _tO&DP~NUFH&KX0C <<-LV?gE9 Uy'QKl^"ny=#;{=/]|R`҅p\bXxso; St5(`6I}ǵlӹ:n$1৛- ~Z x{[\?,a~;0[F̩*f ŔY{!^o(z*BF^̅s/@N+kc}dɩoEK[ڋKۏYZX61/7XNB&iˤS{9oO1i/k[O ΤJk&A#"]&xs[yL![<Ǽ Gc!0kP&].}bݫ 乘i3y(ܙr^g?w֎nu3@n'2n䚯}Pr}/֋=9#.)e乼]y#֜Q7i \NI)f!Xc|\EJ{0/N2>pJ':Ҋڮ666Nۮ[nOέ|Z/lD&{VyJ-t ;B5>@הG,)~ҥ:F㚶:>H^Xb%eD[c]*/hf19ۄypB LLk 6m&/LuSK-%]^ PW~ʫp<0XUn`qxTDiC&k1ߊWI4n; #9mDx܋X/K#Wp,yA4>]K (4㜖7-&g,2ڞf6[-?Ho%^2zr0-uћM@=@w$ aicZ@as}>}GWy@G5"ӯ}_%]+ꮄ|…ɘ}+ߩŘv0WsZZi?m5p<qtm\JTz_@w=SHbG{qovqs4> tƼ8m'r;ڙc^MccyoYi8K<IJN3IA#.Y+i6F|)l3S1Ov6N&XcC,oQall0o?"~%0Hœ1ט`N Ytߣ ˨jzVw*z|7a:>5 s?mW\`_a4 Ga292* vG_Y xLB]*|P[rG@Goگv,"kXM%ՉIԮQ&l6z»Q&Qލ(;|lvғbR |WgPJuvlxG.6Cyf%׎_s1񝐦c74crJkr-Gl-ѭl7aG{7*=f>}G"v+=o&rݸb^,#'zp9:3 \'Y'w<0@&W<{Nxa'c}V'uI 繜ypIuQ~$YVm־)}3yßuP O1nM1?Xǟq8e*Vu7OZ&)l'thQX[{B#LZ3t]|?4Q&2j:,cS[[{bq>.ZV%I~lLΗLWXA Iir"R~,D2RˀwAs'l8Ά,anT틗SYNVZRbקY G秔RuU_je ]B쥵*'6͟>QiqluvdN;n9D|B9>t.-Hr .Y.c"4ƹt3١L|\bB}\_[h IGE̤h !qeXH2X;d|lJS'l,Wrl:ڴ֑VV̪loכxJ6 "7@ OmݬOw$S01W8hwiM;G u}\FZ;-aH9:BvDεDY2hn>OD_H0 n|/untF18f? yhWLC^{DncIpigt?LQNG2wkVt?9ɼt_=s9hyii|N`2:MDiv4uA hE_߮\pqk:;7j3mkjט7O@h?2-! ̺In9mĬ8^WpL krڋq/C`aM8fyܔ82*Ƃgkck!XT ׉`SN7_CSm05Cp%Ǐ_r|+S;aO/f~zF?k:fv-nߘ|31!/ ]?{ۚ&q1z40 i{Ov1mL![IRdk EgLҀ1?"?+iZ6ͫPݛ 7 G%ꙝ[:V|_b [E@5j[]*/*iV[rvn5֛&kߤҎ*&@LXaGqo@۱RBe+E|_p bTGjh !u"ޟ@E {ҹHǺƕ=e'-:{R֞t5$r'Iq|$y=j)ٓfnbO7ź=<=) 'etxo&$!\zwڳ'}0煌 gI=8[sz ArYNuعx]G6k'wy.t ڳl;L0Bl({s#7` `>W8A#>MuFVtoz0gm=[ė9~{68d뷇y#2+ܙc"`3L0>LQ0<ܮ._|MV;- 3>ƍkBl?{ A#f=>Pow>aŷmHm\s:רs4Թ\C5 n3Mgw:ZZ>>zީW'.-[vd~/cטiQ~)=/&=̱~&>G.F;54~2 ~~վľ¶ 2LΛأzYit7ѯʷ:i!SzBO/%/^JtX5aҵcqq؆`a6ra;aMӎ"Y>ƒv|(5oe;~k>8u^=hlc. Ym:9W,轌-"?^l)1oXO/=Ӽ5Cp~7-2cGk|#) vʿ ;%y;as9m,pNluF;N)qgrF"AXU Fܪ~F܉jq?*奖Fq0Nğ즍ǙNbOX0bQ,U=zu!EmK/ oЦ_b8F޾SWQ6,[75xW3~ɹ]e[6Lk[ߋyrOw5ٳB uvIl>u}pj}g[Gy 1k\uֶ5h45Y9'd5opfuya{53x]}VG|zU3`ĝ[,_3ZLhUsxiM<_xf׈Q~Uaml_f^wimuqxr}*j;teRp0(Fߛcюz}k9Yʹ#;0ӵy=:lQ,L0'T󃙞0cFQ{W榐sGWqsL0;^> 3|M0Kw_Da^g?17+x9'yysv=??z!r_G_khWt 3=_Q0| stReWtra^2b)WT`~'04b.h/3 o^>zfOfFkB}ozߕN޾ |o{7 /%ف:޾Z]Tyj籷Kwn1l;:nw{Ծc~a ô"ʙ}Yp|iICk n?= |FZPmu:|D>0leDͿ9&o1߲9Eg1Z3.WH~z,JGI~v0);Yi'H`9"71޸#ލ!cӿ1{uWGGv$ƽ}>]|Q4G:[fҽ1hBJk2]F6ş~˥uv1Cn#&YS= OV_ǩsx9 uh+#mL˴<8>Kb44LV}pz8~9ߏ=E;Gy4Ǽx!=O9/Fwe"I3hMiaJ8(Y5Z^Q:_g4֘4+<cގƣٽk2^? 9`ğ/?HWF:􏏳<7CJxs>՛Z>K(ڮ/[jlTƫCU/]pY5LYk4]}4Gp]R/\>fR‡{*SLba7\-^K>J!\+Sc?-vМmm i,W?&Ox {u:G7w!O,A" B $ B $ B $ B $ B $ B $ BĜ?c,HIENDB`n JGUXiz7PNG  IHDRF%PLTEFnV,f#_! ⇨JKʹU@H  GRQ5lYBv-~Fy:b*dGQ5=0"FLvJPi핲Z]+mIZ!^J O8eA:p~3kVY(s灤âӅ(cV鞹&sZ]vOJ|k N0iUExBTMqz1j/h=rftW8o"_XHzoy%ݏɡ'b?t\ ].gWI{-g%a[Cwc`ep7PTGzD| MhµLbKGDH cmPPJCmp0712Hs IDAThCy\U^1 >P@M DAYdE("TDV}%%Kњi4Mi6McgΙ{<K?wܹw739S1)`RI&L U`hRuGK?u^{ 2;XYY9Xfߥ?&h-swDZѶwS(W45V.=fnc#xMs:xMHY>BFXq=m\2o2r (oX Aڔy\W(;&-PAgk]ڍv_vUGo$]Gh}Z|GgXSmO\Z&eVR e,~c)VVIR5ŏa0=[Xmk~A:&-?Z7 IZ P+3>վߞ=9>'>8NNY ۞qzw|mH\Rj֛ve:Y{ڽ i5=! ea$o8T0~z??Q>aecxPy?5jNNE27aro#%)<(T`0)}%J!sQK2}'<Bղ__>i fbT-Q|kяWǔ^:\iF kvKP-7?EVH}asx:SJ7hX~)G$e@taF4 3}q !_߱@eVb^\MIJ!>e!(SzfZ*Cpa3>$Lj/1[p"$0`%&]r]3}rfƢoF)M Re9t?aCo1kWJ]rY +Hh-%)?n@vhi@F#t&%/Yj*<>FAOi &+p0{ed\ꍙkOXĚP=?R4b*5;fS)v%}%@.q5xNQ]T-9e'PxD|70羬UAՙqe &5ixtE-L)[X!;0r_`e*n)ZZ G[R` *9fHiߊyX";`;婔5x07ԒTU08c# ZV2M\M=ISKB.b$BXob1ꜝX౲uYW J\@}fKjU-ۡn%!AKXnr> P^[{ܼ׶P3W%# vfxM\9 Dg !'Kp.e$9 IPg)C{Ɇ>,)p:.LZU s{&H5)dP.68 guE<3b_<эlTJ{N)S#kmMJt)cQDGuكL@-֚^q?.W{cz (wC$,5~{ԋ\EdNLEQFs[Wrn\hBsbu>z\sn0ܵV? @SGMmU3!B֦̃h\^. jIUA$rw<ڱ/V(:P3JBAF~S7uTFA (g?ӰE8YRnU<5^w=j U:s=zt}~y{Hwl ~iw#$pno_vlTȨadRI&L )`uPhIENDB`FFMcBԯ0JFIFddDuckyP&Adobed Oi       (0@!2 1"&!1AQa"2qBRb# @30C$@ !1Aa!1AQaq @0 JUSG78zvϩA1rˉc.$ɯDIZӶʕչ{駺9RNՋVŘv6\zVȭ3!Yy<3MŵOghDNdLt\}mcLL!Y|6Z *峲(Ge̲ڮb,uC/3!m:,W_e&dxM|3:ʟ q,J֟ہ rw{KO'|~^տO+݌Z뙗6Ŵ+ysQȭW3wmk1Y5O/rTDt ڈN~ )LMDt.6suEXEdXd3G31c tqTmPW0_qxLJk{}dղb6i5#cYGgN2aI,66X0D5Іj0F 1RpMjD5,D >,o4 #cg!Gl6F7?K5x <32Q23(J Čl261 #Jn|sc1Y3~F⒳. F=05"`d`j5~?o$NF+>!TH8ː2F|O!TH4.@7e?&3`InYȩM⵳>39:*P1$M;ѷCڷz8!l{3w1$z`\dϳuZ)84˸*Wi`=Nr>>YL}7|ۑkƜ`4:䤫FyڕX܎6&z1}C-TT❣HG#wqUEǔ^x,Yf:Uv&g VV㣎1!hUjGld+s=cx,Ѳ+wG*)=puͳ ,|jP-a VvɐgcTAnK(Zf?ME\랹B8i&31-ݵn*3GJ(!Uc5rfSRpOr%ꄫ}ٌO!4 b^`zm+~xc?`帻 Mݲ/Eb<:j:,?ǜsQTҚ,YMmTaBJžow5XQu {*j*V#o mMr gln+|O,Mɿʟ))Wkyپ6/b85-% 9%۷;|3ԢP5{jYAz;oǗ}R~qz#HufM{q pp}$rKZH\[j2I`R9o nbXmz|S~ӶRFE7>= Ն\њYfMshnNs~/n} i:k)Ŷ.]R2\=?B]>2݅z=)i;q1VW_.Lx=1YpF)'2/.lR)5?!CY1?p $$wzqcjuʟf^η-a:N?#3m.py)c_lߖu'ddo7q7/>u=x"n)4@ѷD F=bܑߕZ{@d̟*?=i@EŠX=PA$9CׯX(t#|CMAN(o+}5OtoBlU~rFhSaqe1ScyMs.CY6aE͡_=qx3gۦ.?k;a 7}W#H^_?ͨnͰwkYvj5•y4 r lKY0D&}ɟ@EP{χh!rÄu/ق^3RH *_&iXC&_@MAzڐ':2H8 oG_RN>R H4~rpq)>^Pc\&M]E?ո0G@?eL^7X 2kb4'o ]|)U\\~N0A*^c/HCa'GbfM8y.rqEݞ8:߽z% 8I( P . Dhttp://www.sickkids.ca/default.aspDhttp://www.sickkids.ca/default.aspDhttp://www.sickkids.ca/default.aspDhttp://www.sickkids.ca/default.aspv,http://www.utoronto.ca.http://www.utoronto.ca/l)Equation Equation.30,Microsoft Equation 3.05Document Word.Document.80.Microsoft Word DocumentJ/ 0|DArialgsRomanTT-ܖ 0ܖDTimes New RomanTT-ܖ 0ܖ DWingdingsRomanTT-ܖ 0ܖ0DTahomagsRomanTT-ܖ 0ܖ"@DSimSungsRomanTT-ܖ 0ܖ@0.  @n?" dd@  @@`` 8 F1q            5")< = M    !_ "& =4 #    )       = / (,     J S 16  g ZI ob  Y!KRC  ;L?2$hiALвO"$EHX9$ b$k6LH(v\n  "$$$$$$$$$$$b$JGUXiz7 ,$$$$$$$$$$$$$$$$$$$R$FMcBԯ09 0AA r3ff@f8125Sʚ;?8ʚ;g4KdKddm 0\ppp@ <4dddd w 0T- <4!d!d x 0T<4BdBd x 0T___PPT10DSimSunew RomanTTx 0xDTimes New RomanTTx 0x ___PPT9z~0>TbW^_UH I kV*? O  =,B NSerum Diagnosis of Chronic Fatigue Syndrome (CFS) Using Array-based ProteomicsOO COutline$Objectives Data Set Methods & Results 3.1  Preprocessing (identification of biomarkers) 3.2  Classification model Conclusions^'fNvf f'"N "> Objectives  $f;Identify biomarkers for CFS/CFS-like diseases using SELDI-TOF MS technology Evaluate performance of the identified biomarkers to distinguish patients with CFS/CFS-like from healthy people Determine the best experimental protocol for large sample studies by choosing the best Fraction/Chip/Laser Energy combinations f<ffffDData Set   fTData Analysis Pipeline fPreprocessing Baseline subtraction (already done) Trimming low m/z values Normalization Peak finding and alignment Quantification of aligned peaks Merging replicate samples Classification Do a 10-fold cross-validation (CV) For each step of CV Split samples of preprocessed data into training and test sets Perform biomarker selection on the training set using t-tests Built prediction model on the training set kernel-based K-nearest neighbor (KNN) classifier Evaluate performance using test set FfPfPfP7fPvfP1vfP%vfPfPfP f}7  1  %?"WTrimming low m/z values#f  Low laser energy allows peaks in the low mass range to be well-visualized High laser energy improves visualization of peaks in the high mass range Many studies (e.g. Baggerly et al. 2003) indicated that there is a noisy m/z region near the lower limit where the machine can not record stably. For the above reasons, we trimmed low m/z values using the following thresholds: For low laser energy condition, we trimmed low m/z values less than 100 For high laser energy condition, we trimmed low m/z values less than 2000 hxfZfZfZxffb.lXG*^Global Normalization (Li 2005)&f Given a spectrum with intensities Xi (i=1,..,n) for all n m/z values, normalized intensities Xinorm can be computed by Xinorm = s*Xi where s = (median of the total intensities among all spectra) (the total intensity of the current spectrum). Multiplying raw intensities by the factor s equalizes the median (mean) of the total intensity among compared spectra.wfrfH fvf#94ff3J*fK>: 4 8jPeak finding -- Why2ff6h)Peak finding-- Algorithm (Tuszynski 2006)>* f /bPeak Alignment -- Why$f 7iAlignment  Algorithm ------Maximal cliques & real representations (Li, 2005, Gentleman 2001 ) f`  f 'f   +_.Quantify aligned peaks for individual spectra (/-f#fEach aligned peak location (m/z) can be treated as an interval, m/z * (1-0.2%,1+0.2%) The intensity of each of the aligned peak location of individual spectra can be quantified by the maxima in the interval.3,! UMerging replicate samplesfAfter quantification of the aligned peaks in all individual spectra, we averaged the intensities of the two replicates for each samples. The averaged intensities were used to build our prediction model.fH. Predictor --K-Nearest Neighbor (KNN) MethodD/ f"fTo classify a new input vector (observation) v, examine the k-closest training data points to v and assign the object to the most frequently occurring class Neighborhood is defined based on a mathematical distance measure Deficiencies: The individual points in a neighborhood may have very different similarities to v (distances from v), but they all have the same influence on the predictionfPfP.f0f?QPff:IG Predictor --Kernel-based KNN Method (Hechenbichler and Schliep 2004)dHf ff!f,( To classify a new observation v, examine the k+1 nearest neighbors to v according to Euclidean distance (d) The (k+1)st neighbour is used for standardization of the k smallest distance by D(i)=D(v, v(i))= d(v, v(i))/ d(v, v(k+1)), i=1,& , k Transform the normalized distance D(i) using a Gaussian kernel function K(.) into a weight w(i)=K(D(i)) Assign a prediction label to v based on where y can be either CFS/CFS-like (r=1) or NON-CFS (r=0) disease lfPfPfPfPfPJfPfPf'ff  E*fffUfy <76z2e  0cZThe number of biomarkers identified in each condition, and the number significant (p<0.05)[[fZ  3fThe number of biomarkers identified in each condition, and the number significant (p<0.05)  Cont.bbfa 4gComments  +f 1dZPerformance of the kernel-based KNN predictors using selected biomarkers in each condition[[f(]ZBiomarkers used in building prediction model for condition: H50, Low laser energy, and F6[[f  Conclusions  (fdBased on our analysis, the best combination (laser energy, chip and fractions) appears to be Low laser energy/H50/Fraction 6 We identified 9 significantly expressed biomarkers (p-value<=0.05), which are located in the 3 m/z value ranges: 499-503 526-528 7784-7785 Using 14 biomarkers identified from the combination, our predictor can reach ~80% accuracy.^f fqfvf\f^ fq\9k Limitations  For all combinations of experimental protocol, we used the same m/z shift interval (m/z*(1-0.2%,1+0.2%). A better choice may be obtained by estimating it for each combination from QC samples We did not take the multiple testing issue into account in this analysisf,@!VAcknowledgements  f.We used following R packages to perform the analysis in this study caMassClass (Jarek Tuszynski) PROcess (Xiaochun Li) kknn (Klaus Hechenbichler and Klaus Schliep) This research was supported by funding from Ontario Genomics Institute and Genome Canada, through the Centre for Applied Genomics. DfbffffDbD    /   0` Ot{h______` M <ff33̙3` +ffO=ff̙H7` fff3f̙` Tff33ff` 0Ky{kOz` )R{f` GiIfff̙fR` ̙|̙3f` 3ff~>?" dd@'?lFd@vlnK'o`P( n?" dd@   @@``PT     o (`0p>>  4,(  , , 6T " `0  X Click to edit Master title style!!  ( , 0Ƞ " `  RClick to edit Master text styles Second level Third level Fourth level Fifth level!    S   , 0 "` `  Z*    , 0̬ "`   T*     , 0< "`P  T*    T I +, "c} h (, s *"3 ), 6X"I = .>" "2;"  0 *, S BAnewlogoSick Kids"@  0 4, S nA#:uoftbuttonUniversity of Toronto button"`  0H , 0޽h ? 3ff~___PPT10i. s+D='  = @B + Network 1 0  ((0(  0ZB 0 c $D"  0 6p  "&wf   X Click to edit Master title style!!   0 0  "wQ    [#Click to edit Master subtitle style$$   0 0  "` `   Z*    0 0  "`    T*     0 0\  "` `   T*    T `]  0 "]p h2  0 s *"`]h2  0 s *"]h2  0 s *"]Eh2  0 s *"`h2  0 s *"h2 0 s *"Eh2 0 s *"yh2 0 s *"`B h2 0 s *"B h2 0 s *"EB h2 0 s *"yB n2 0 0",B h2 0 s *"`u  h2 0 s *"u  h2 0 s *"u E n2 0 0"yu  h2 0 s *"`(  h2 0 s *"(  n2 0 0"( E n2 0 0"y(  h2 0 s *",(  h2 0 s *"` Z n2 0 0" Z n2  0 0" EZ h2 !0 s *"y Z n2 "0 0"`  n2 #0 0"  h2 $0 s *" E h2 %0 s *"y  h2 &0 s *"A  h2 '0 s *"yA  `B (0 s *D"H 0 0޽h ? 3ff~___PPT10i. s+D='  = @B +1 0 zrp  (      0tj  P   j  P*     0yj     j  R*  d   c $ ?  j    0|j   0 j  RClick to edit Master text styles Second level Third level Fourth level Fifth level!     S   6j  _P  j  P*     6̆j  _  j  R*  H   0޽h ? 3380___PPT10.;pʽ0 0P(    0  P     D*0   0       F*0   6`$  `P    D*0   6(  `    F*0 H  0޽h ? ̙3380___PPT10.J0@@ 1 0 WO0(  r  S <9j ,``  j  5  <Bj P` +Pingzhao Hu W Le, S Lim, B Xing, CMT Greenwood and J Beyene Hospital for Sick Children Research Institute and University of Toronto The Sixth International Conference for the Critical Assessment of Microarray Data Analysis (CAMDA 2006) Duke University Durham, NC, U.S.A June 8-9, 2006 N,0 f2Nf,6 TH  0޽h ? 3ff~___PPT10i.ÜJ+D='  = @B + 1 0 @(  x  c $ȴj , `p  j    S ,j ,  j  "+s,H  0޽h ? 3ff~___PPT10i.ʜpX+D='  = @B + 1 0 6(  ~  s *j ,`   j  x  c $pj , j  H  0޽h ? 3ff~___PPT10i.6+D='  = @B + 1 0 *^.(  x  c $j ,` `  j    `,  ^ #"."    *@ `  j  / <j ?P& `,  M864o( . <j ?@ & P,  N3072o( - <(j ?0& @ ,  N2976o( , <Lj ? & 0,  W Total spectrao( + <D ?P`&  T High, Low   o( ) < ?0P&  S High, Low  o( ( <d# ? 0&  V Laser Energy  o( ' <+B#style.visibility<*w%(D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*w%(D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*%(D' =%(Dh' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*1%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*1%(+  1 0 aYt(    s \j 0e0e ,#" 0e `  j    0Di   `r$D 0>6___PPT9 MThe height of peak intensities at certain m/z values indicates the presence and the approximate amount of corresponding proteins or peptides in the sample However, not all peaks at a m/z value are related to a protein or even a part of a protein We need to search for those peaks that may represent a protein or a part of a protein *N df2M,*p `@P`pH  0޽h ? 3ff~___PPT10.V|+|3D'  = @B Du' = @BA?%,( < +O%,( < +D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*%(D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*%(D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*L%(+% 1 0 <4p(   0 NA ? ?,0`     s a 0e0e ,#" 0e `   H  0޽h ? 3ff~___PPT10i.V|+D='  = @B + 1 0 OGX (    c $Ԃ , `   ( @` + 0  ( 2 > s  0e0e ,#" 0e0    E 0  @0~$D 0JB___PPT9$ Assume two peaks: R1 at m/z value L1, and R2 at m/z value L2 are detected in two spectra, respectively. It is known that the m/z value of the same peak in different spectra may have a small shift (0.1%-0.3%). The shift must be adjusted so that peaks in the given shift interval (say, m/z *(1-0.2%,1+0.2%)) are aligned to have the same m/z value The objective of alignment is to estimate common m/z value L3 of the peaks in the given shift interval across spectra F df2 df2tJ18Dp `@P`p8   Xp@   G @  D TB  c $DTB  c $DPP`R  0 PP`b   0P`R   0`b  0  < 0\ :R1  <t `G :R2TB @ c $DTB A c $DP PP B 0 `p  :L20 2  C 0T m :L10 2 TB F c $D` ` P W Bt 0 p  8L30 H  0޽h ? 33J B ___PPT10" .0f+B(D '  = @B D ' = @BA?%,( < +O%,( < +D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*Eh%(D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*Eh%(D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*EZ%(D4' =%(D' =%(D' =4@BBBB%(D' =1:Bvisible*o3>+B#style.visibility<*EZ%(+ 1 0 9: _(   ~   s *0 ,` `      c $ ,@#   ( @` +  0  ( 2 ,  0H P  ?Find maximal cliques: {1,2}, {3,4,5,6}, {7,8}, {8,9} Real representations: Find common m/z region for each maximal clique and estimate the aligned peak centers using maximum likelihood estimation (MLE)" f2Wp8X8  : P TB   c $DTB   c $DPPTB   c $D  TB   c $D  `R   0 PP`b   0P`R   0`R   0` ` `R   0 P `R   0 0 `R   0P`R   0`R   0 P `R   0 `  `b   0 ` P `b   0 0P `b   0 P`b   0@ ` p `b   0 P `b   0  `b   0`b   0 P    <8 0\ :R1   <H p  :R6   < 0   :R8   <f 0   :R5   <h < w :R4   < `G :R2   < w :R7 !  <0  :R3 "  <    :R9ZB #  s *D ZB $  s *D  PZB %  s *D` ` ZB &  s *D  ZB '  s *Df3  ZB (  s *Df3P P ZB )  s *DP ZB *  s *D0P0 ZB -  s *DԔ PPZB .  s *DԔ` ZB /  s *DԔ 0 ZB 0  s *Df3Ԕ P ZB 1  s *D ZB 2  s *D ZB 3  s *D ZB 4  s *DPh2 5  s *"` @ZB 6 B s *D   7  0  @' NNot a maximum clique0 2f 9  04 `    PAligned peak centers 0 2fH   0޽h ? 33___PPT10i.0f+D='  = @B + 1 0  6(  x  c $ , `p   x  c $` , `      05 0 @ 2Black: raw m/z values with intensities Red: Aligned peak m/z value. It has no associated peak in the raw data Blue: left and right intervals of the aligned peakB0 2'fD/, +e8  p`   p@ TB  c $D` p` ZB  s *D ` ZB  s *D0  0 ` ZB  s *DP P ` ZB   s *D ` ZB   s *D ` ZB   s *D ` h2  s *"`  ZB B s *D P   0`$    OThe intensity is quantified as the intensity of the aligned peak location (red)4P0 2KfH  0޽h ? 3ff~___PPT10i.U+D='  = @B +} 1 0 e$(  r  S  ,0   r  S  ,0P    H  0޽h ? 3ff~___PPT10i.+D='  = @B +  1 0 G?0008(  8~ 8 s *Ǽ ,    8 s *μ ,`PP   ( @`wF   8  TB 8 c $D@@ TB 8 c $D 0 `2 8 0`0 ` `2 8 0  P0 `2  8 0P@ p `2  8 0 0 P ` `2  8 00 ` `2  8 0` `2  8 0 p `2 8 0@`2 8 0 `2 8 0 `2 8 0 0 `2 8 0 P `2 8 0 P `2 8 0 ` `2 8 0 0 ` Z2 8 s *@ p Z2 8 s *  @ Z2 8 s * P Z2 8 s * `  Z2 8 s * @ P p Z2 8 s *  Z2 8 s *` Z2 8 s *0 ` Z2 8 s * 0 Z2 8 s *@pZ2  8 s * 0 ` Z2 !8 s *0 @` Z2 "8 s *  Z2 #8 s *P p kN     $8    Z2 %8 s *  &8 <yi     7x0Z2 '8 s * P  Z2 (8 s *p 0 ` N  :  )8  : f2 *8 6Ԕ ` `  +8 <Qi ' : 9k=10ZB ,8B s *D0 p ` N    -8   f2 .8 6Ԕ   /8 <|    9k=60ZB 08B s *D H 8 0޽h ? ̙33ffy___PPT10Y+D='  = @B + 1 0 i5<(  <~ < s *L ,    < s * ,`P    ( @`^ 2< 6p 1< c &A l??"?p @  l 3< 6P   ,k is implicitly hidden in the weights - if k is too large, k is adjusted to a smaller value automatically, since only small number of neighbors with large weights dominate the other neighbors (very small weight-no influence on the prediction)0 2 5< 00i 0` QWe set k=3 in the study0 2fH < 0޽h ? ̙33ffy___PPT10Y+D='  = @B + 1 0 .(    0  0 EResults"0(2$f  0i    I#We first define some concepts used in the section Condition: Experimental protocol (Fraction/Chip/Laser energy) Biomarkers: Here we mean they are the aligned peaks Differentially expressed biomarkers: Aligned peaks that have p-values less than 0.05 selected by t-test.  2 f2" f21 f2 f22 .fff  f*#HH  0޽h ? 3ff~___PPT10i.`mb+D='  = @B +$ 1 0 ##b"J'#(  x  c $( ,` `@   x   J #"*}# f ? <3 ? G  z0___PPT10:___PPT9 h24,(" dc  @@``f = <7 ?  G z0___PPT10:___PPT9 h52,(" dc  @@``f ; <DE ? X z0___PPT10:___PPT9 h29,(" dc  @@``f 9 <XH ? Xz0___PPT10:___PPT9 h10,(" dc  @@``f 7 <X ? z0___PPT10:___PPT9 h40,(" dc  @@`` 5 <b ? z0___PPT10:___PPT9 3# of differentially expressed biomarkers (p<0.05),4(" d4c  @@``d . <k ?  G z0___PPT10:___PPT9 f1220((" d"  @@`` , <8v ? G z0___PPT10:___PPT9 CM10 High Stringency-F5 6 " d#"    @@``t  < ? G z0___PPT10:___PPT9 v10538(" d#"  @@``  < f ?G z0___PPT10:___PPT9 CM10 High Stringency-F66 " d#"    @@``s   <l ? X  z0___PPT10:___PPT9 u3158(" d#"  @@``   < ?X  z0___PPT10:___PPT9 CM10 High Stringency-F46 " d#"    @@``s  < ?  Xz0___PPT10:___PPT9 u6138(" d#"  @@``  <L ? Xz0___PPT10:___PPT9 IMAC-F56 " d#"    @@``t  <D ?  z0___PPT10:___PPT9 v 2358(" d#"  @@``t  < ? z0___PPT10:___PPT9 vH50-F46 " d#"  @@``  <` ?  z0___PPT10:___PPT9 # of biomarkers identified8(" d#"  @@``|  < ? z0___PPT10:___PPT9 ~ Condition1: " d #+  @@``ZB  s *1 ?ZB  s *1 ?  ZB  s *1 ? ZB  s *1 ? ZB   s *1 ?ZB ! s *1 ?  ZB # s *1 ?ZB $ s *1 ?XXZB % s *1 ?  TB - c $1 ?G G TB 6 c $1 ?   & << 04___PPT10 B___PPT9$ k1Only conditions (total 32 conditions) with at least 2 differentially expressed peaks (p<=0.05) are listed Zk fif/f$f + 0 ` ` KHigh Laser Energy0 2fH  0޽h ? 3ff~___PPT10i.N[+D='  = @B +_R 1 0 vQnQ0DQ(  ~  s * , `   O @   #"2&p@   d  Hi ?|  z0___PPT10:___PPT9 Z( " dc  @@``|  H$  ?"`p | z0___PPT10:___PPT9 d36( " dc  @@``m  H$ ? p z0___PPT10:___PPT9 c9( " dc  @@``m  HD0 ?  z0___PPT10:___PPT9 c3( " dc  @@``n  H: ? z0___PPT10:___PPT9 d40( " dc  @@``{  HD  ?"`z0___PPT10:___PPT9 c4( " dc  @@``m  HDO ?z0___PPT10:___PPT9 c9( " dc  @@``s  HlV ?@z0___PPT10:___PPT9 i#p<0.05( " dc  @@``{ ~ H  ?"`| z0___PPT10:___PPT9 c4( " dc  @@``{ | Hh  ?"`p | z0___PPT10:___PPT9 c7( " dc  @@``n z Hi ? p z0___PPT10:___PPT9 d12( " dc  @@``m x Ht ? z0___PPT10:___PPT9 c5( " dc  @@``n v H ? z0___PPT10:___PPT9 d12( " dc  @@``n t HXu ? z0___PPT10:___PPT9 d20( " dc  @@``m r H ? z0___PPT10:___PPT9 c9( " dc  @@``t p H؋ ?@ z0___PPT10:___PPT9 j# p<0.05( " d c  @@``} G HP  ?"` p | z0___PPT10:___PPT9 e449( " dc  @@`` E H  ?"` p | z0___PPT10:___PPT9 zIMAC-F5 $ " d "    @@``} C H  ?"`0p | z0___PPT10:___PPT9 e418( " dc  @@`` A H  ?"`p 0| z0___PPT10:___PPT9 yIMAC-F4$ " d"    @@``o 8 H ?   z0___PPT10:___PPT9 e376( " dc  @@`` 6 H ?  z0___PPT10:___PPT9 CM10 Low Stringency-F4& " d#    @@``o 4 H ?0  z0___PPT10:___PPT9 e328( " dc  @@`` 2 H  ?"` 0 z0___PPT10:___PPT9 CM10 Low Stringency-F32 " d#"    @@``o / H ?  p z0___PPT10:___PPT9 e391( " dc  @@`` - H ? p z0___PPT10:___PPT9 yIMAC-F3$ " d"    @@``o + H ?0 p z0___PPT10:___PPT9 e370( " dc  @@`` ) H  ?"` 0p z0___PPT10:___PPT9 CM10 Low Stringency-F52 " d#"    @@```  HP ? |  z0___PPT10:___PPT9 V$ " d"  @@``t  H ? |  z0___PPT10:___PPT9 j$ " d"    @@``}  H`,  ?"`0|  z0___PPT10:___PPT9 e341( " dc  @@``  H>  ?"`| 0 z0___PPT10:___PPT9 zIMAC-F6 $ " d "    @@``o  HTH ?  z0___PPT10:___PPT9 e706( " dc  @@``   HXR ?   z0___PPT10:___PPT9 CM10 High Stringency-F62 " d#"    @@``o   Hh\ ?0 z0___PPT10:___PPT9 e809( " dc  @@``   Hf ?0 z0___PPT10:___PPT9 CM10 High Stringency-F52 " d#"    @@``}   Hp  ?"` z0___PPT10:___PPT9 e409( " dc  @@``   Ht{  ?"`  z0___PPT10:___PPT9 CM10 High Stringency-F42 " d#"    @@``o  HL ?0z0___PPT10:___PPT9 e417( " dc  @@``  H ?0z0___PPT10:___PPT9 CM10 High Stringency-F32 " d#"    @@``o  H ? z0___PPT10:___PPT9 e299( " dc  @@``|  H ?  z0___PPT10:___PPT9 rH50-F62 " d#"  @@``o  H ?0z0___PPT10:___PPT9 e351( " dc  @@``|  H ?0z0___PPT10:___PPT9 rH50-F42 " d#"  @@``|  H` ? @z0___PPT10:___PPT9 r# of biomarkers( " dc  @@``  HD ? @ z0___PPT10:___PPT9 u Condition2 " d #"  @@``|  H< ?0@z0___PPT10:___PPT9 r# of biomarkers( " dc  @@``  HT ?@0z0___PPT10:___PPT9 u Condition2 " d #"  @@``ZB  s *1 ?@@ZB  s *1 ?  ZB  s *1 ?@ ZB  s *1 ?@ ZB  s *1 ? @ ZB   s *1 ?ZB ! s *1 ?0@0 ZB " s *1 ? @  ZB # s *1 ?ZB $ s *1 ?ZB % s *1 ?  ZB * s *1 ?p p TB 3 c $1 ?  ZB B s *1 ?| | TB q c $1 ?@ TB  c $1 ?@  ' 0X 0 0 JLow Laser Energy0 2fH  0޽h ? 3ff~___PPT10i.N[+D='  = @B + 1 0 m(    0 P  b(Using Low laser energy, there are 13 conditions (Fraction/Chip) that identified at least two differentially expressed peaks/biomarkers (p<=0.05) Using High laser energy, there are only 5 conditions (Fraction/Chip) that identified at least two differentially expressed peaks/biomarkers (p<=0.05) )0 2f f ff]ff ffW  H  , `    H  0޽h ? 3ff~___PPT10i.MSL+D='  = @B +M 1 0  MM eBBL(  ~  s * , `p    AI  `@ # #">2 3 `   d  <i ?` `P z0___PPT10:___PPT9 f13* " d&  @@``f  <d& ?0` P z0___PPT10:___PPT9 h76.1* " d&  @@``f  <) ? ` 0P z0___PPT10:___PPT9 h69.8* " d&  @@``  <0 ? ` P z0___PPT10:___PPT9 )Low_Laser_Energy_CM10_High_ Stringency-F3,* " d*')   @@``m  <@  ?"`` ` z0___PPT10:___PPT9 a2& " d"  @@``p   <|K  ?"`0`  z0___PPT10:___PPT9 d67.9& " d"  @@``   <U  ?"` ` 0 z0___PPT10:___PPT9 t60.36 " d#"  @@``   <8V  ?"` ` z0___PPT10:___PPT9 (Low_Laser_Energy_CM10_Low_ Stringency-F36) " d(#" )   @@```   <t] ?p `` z0___PPT10:___PPT9 b15& " d"  @@``b   <O ?0p ` z0___PPT10:___PPT9 d59.6& " d"  @@``b  <v ? p 0` z0___PPT10:___PPT9 d60.3& " d"