From 5e58c4f351995a1d66f06af3ec136a8512754891 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Irene=20L=C3=B3pez?=
 <45119610+ireneisdoomed@users.noreply.github.com>
Date: Wed, 13 Mar 2024 09:41:18 +0000
Subject: [PATCH] test(coloc): add coloc semantic test (#538)

* test(coloc): define fixtures and parametrise coloc tests

* test(coloc): compare dfs with assert_frame_equal
---
 tests/gentropy/conftest.py                    |  44 -------
 .../coloc_test_data.snappy.parquet            | Bin 11927 -> 0 bytes
 .../method/test_colocalisation_method.py      | 119 +++++++++++++-----
 3 files changed, 85 insertions(+), 78 deletions(-)
 delete mode 100644 tests/gentropy/data_samples/coloc_test_data.snappy.parquet

diff --git a/tests/gentropy/conftest.py b/tests/gentropy/conftest.py
index 13992f8f2..78bd567da 100644
--- a/tests/gentropy/conftest.py
+++ b/tests/gentropy/conftest.py
@@ -3,7 +3,6 @@
 from __future__ import annotations
 
 from pathlib import Path
-from typing import Any
 
 import dbldatagen as dg
 import hail as hl
@@ -648,46 +647,3 @@ def sample_data_for_susie_inf() -> list[np.ndarray]:
     lbf_moments = np.loadtxt("tests/gentropy/data_samples/01_test_lbf_moments.csv")
     lbf_mle = np.loadtxt("tests/gentropy/data_samples/01_test_lbf_mle.csv")
     return [ld, z, lbf_moments, lbf_mle]
-
-
-@pytest.fixture()
-def sample_data_for_coloc(spark: SparkSession) -> list[Any]:
-    """Sample data for Coloc tests."""
-    overlap_df = spark.read.parquet(
-        "tests/gentropy/data_samples/coloc_test_data.snappy.parquet"
-    )
-    expected_df = spark.createDataFrame(
-        [
-            {
-                "h0": 1.3769995397857477e-18,
-                "h1": 2.937336451601565e-10,
-                "h2": 8.593226431647826e-12,
-                "h3": 8.338916748775843e-4,
-                "h4": 0.9991661080227981,
-            }
-        ]
-    )
-    single_snp_coloc = spark.createDataFrame(
-        [
-            {
-                "leftStudyLocusId": 1,
-                "rightStudyLocusId": 2,
-                "chromosome": "1",
-                "tagVariantId": "snp",
-                "left_logBF": 10.3,
-                "right_logBF": 10.5,
-            }
-        ]
-    )
-    expected_single_snp_coloc = spark.createDataFrame(
-        [
-            {
-                "h0": 9.254841951638903e-5,
-                "h1": 2.7517068829182966e-4,
-                "h2": 3.3609423764447284e-4,
-                "h3": 9.254841952564387e-13,
-                "h4": 0.9992961866536217,
-            }
-        ]
-    )
-    return [overlap_df, expected_df, single_snp_coloc, expected_single_snp_coloc]
diff --git a/tests/gentropy/data_samples/coloc_test_data.snappy.parquet b/tests/gentropy/data_samples/coloc_test_data.snappy.parquet
deleted file mode 100644
index 71b3913ebfc68e1bc5e0425b3915d59e66406684..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 11927
zcmbVy1z43!)b0j0DGi&J?(Pzt1uBA+M-gm9N)9Pti^m{TPz1$5L2LyTF|dH4B%~Xp
zyOEHTxF0G<|5NwBcRzbQYvP@?X4X41-=5judZP_$6f%XK!c7_NywSpHK%u6PNhAg#
zD$@54g+!sFFi}3<i!WgmqV&)cU<L{UiA2Ld`j%O2SrBynz@H7N|CbG!D6Evv7p~1N
z`_Yn?ebJ@bfBPd*CJZR?%PA4<gKw$6b!V}AM0ef`tI`n7d+4ba{rjm=u`v6osgh|~
zNU9tRBsCUxk~#|mNrMngLbM3cCPar2T|)F2?O4`w*i+b9><H|?3JW5vJSD`Hq|U)d
z;MD(FSab+ac0Ywf^w-?aG>Zz+8@vB09I}5cVX<S8A$n~0ci~X|pQS7!#Hj54SvYim
z*0AsrBe(l^;ehY8ENnz{?EYIgObLXZSuRO$QUAA)_4J%QNb1fM%Vd(eCz1MRA*-u0
z?pUhyq+Lo8T8X4?*+fV<AwW`B8X@F1AsZJlo;Cz4$sDc}KDlMTwDBOT6Qp@AD!JVp
zI{)l)5l>bpNc*AWB7U{-*ECt3pluOJD7luu)_%98#i44(PF5#)TLcn{MdZI+6TB_r
z2*twt@3<vbWOaf+v!bQ!e`&HhL805f#`@Q$WOY3PsNw8kX;0E1Izb3oL)9&RDYK-B
zLOJuLjOEj%)IVvm2C<yJo3fny3qXscVcA8<6+%o&8cNfI#FI4$mOL#;L_3Z&4sVL6
zo8eM3vWCVt9tmZ~qVnD7Qr)kbtnr<lMdfDjYt2%ctU<8jW=W`TQD0Qw*!eN&pRn(&
zX#7;)$oUIw3BIJs8U#6uf&8y7>atu}jP7@C{sXyWg{-m2jV8ejv4atYe+XGqHD7)y
zv#5yz#+OS4mc-uj6Z_XRS@SzL1jO?GUn>{$WKDH5_8*CFfzl+nu^j&gs3A$ya)^+d
zgsdloimXYHwAf)aCp8L34#xqC>^JT-HJ22zsr|icWX<oyXnrSVQGFxkUq(UJ{Lajx
z`o_$^;Ffw!*8I-Q{{{Aa$ls~?UqKhC(Q@`sN@CDbRb^!3$zdUB@i;J%wDM<3T5eZ}
zja!|h%_vXOVWcPNGSV~JFbWcRi+>gtwVySN+(a9Te-{?r@3o99gk6jO78b;4As0nF
z>MZ_E4D@sspP_zwy5E%^@hmQ0NvT~)=@^b2Wcl@R`#|K&eYPJDT&fp$_3wY3L09fG
zY?8S)j~Mcl@|B*5fkK+k;4Zcd{OX$4y7ml4&^Php+>@>g5so{|I6Jt2b~ZJt&&~i$
z#fRvw-mSn&$B;n7<669~D=jZ4TNSpfc^+ssp#}{KMO!D6>0prO(7A2u?C`0T&8Rj)
z5PX;qe@=O-0lMq_tKSMM0|OooxYmi0+2|(LbG3@FA$mGbE=?Wg*CxMAV4{O`r{0$J
z(K0~l(&9!r0^sj=TS4M;AIdRFG4H&p3LBzAPKLP2!=Y6Z0!kMI;qV_F4|qkW!K-Y-
z`Am;GjBQ$LESsbVYhu#mf0LI1I+x-#_l?T2!-lHkiSHNCR@-|Qo~`7ECe9YT&X)rD
zjGYB(4*XCOUTnd=wh~Vq7|2ebmV|KM+Na0KN}$|#__t5ib$G`MyM4ap?BLk*Bx(~b
z$4Pk|n`*s8!B}wF{GJ2_sClbYpcO0&-iGVPg8M{4gYLS#aCJWB=@h)oZz2w--r1xC
zFG#_;LDe19v<Y|z&0v)IO&$mvrlYfzC&ulNuTU{xj#HS(die#iko2JmAtoj`NT)YG
z>?sZB?p2r63(~-pfedM%s~{|Vsb24@E)A(SG>d0L8Ns&lgVK%9W2kOorSXw>g}5dy
zr<XUW32{0;ET-0%0&N*u<>ythaIN=)|K)gbh!+@`yBNR>1-V+vnT-q(Df7vEN0$hc
z_XngbJZ?t`n>?xVR?EQVfrzKwJSDigR6ERLa}lEY5R#Z|%>Y+VrQK-QC=BizD++4o
z*?@lCmU{-O9q3eTsK@g+?MNwylIW8`f*QJOTg!P{(XEjF<3m225ZdBsxAkiey8qB#
zWn88OrH{?6*Ip$FLQTh$;+OG3&W`$&cdUXia=WSC=Jz<nU8OGkQg0Teh^&_PbmInS
z;(x9XH;&#M<tdSWCk=-q9yuMW9z~k>xB~a<2!nIk#&h1f7x7l(P%DRA5!lt$6_+F-
z30no$`3ULo!={rl2I1#l<C_lsUe`CsfRJI8Wq<t$%Ac}lMD?TS&@-x{>~dOoA96b6
z_Wl}_Ao6OZ6<G$FCx*^5Mu|aKhRv?<=bS)w=&C_}C@16;IGwsV#R*$HqYTe27XiOV
z)(14Rvax)n>t>^^JTS6io6*@<WYAg8==e;H1H5l+eLqv2jTiPDbgg+M3Z7yKGM?M{
z;SB`Z8U!VwL638JNP8Iwo^2SpHBJwY`6cgl*rcKh-mcHJ|ENNTN=ImpiR9vQ7wcsm
z)C#e*RGi-0BnF_eQgU|55QU<|n*vWYx#5gohmm~0H2ADEy3APc7Rzgi^`0c+51@EI
z;wmF>C)=@=-I_+qyV7|LD<mQEkMsbAo^*7$c8+fwR3hp^a@Ec{I@q?$>W>a{TIlaU
zSZ2KhaLw6_^3n~WW8<<pn6(CLtMJ^Pf6NJycY-}Frt=UDKB9WSc@}LLJhg`1u@bQ~
zzRc3h6opNjE=Bt2QvmP!C||hgC?W+a<Y^o$!$E@K9S8gQAjYUtt}?3<Ka*3Klvwit
z&DffsvJqtkH}i<TC%WvQpZ@CPq31#%yNQnF;{#e)SK)Lx+MEik@0$AL-KM}>Q9i4V
zGA?*&7(Cp-$`0;-n2imcA4a<;-5)bPDMd5}2aQ}llq1{IZ$f1y=1>vOrvB@ly(nYn
z*ZC_unc?9ZWZl3uhDuWTkCd1VqvM^FOT%PtP$-RR9amujrJnrEhU|G{v^&nE*PRQ7
zR3F~-Sl|XZrHfaSt;P|aTR*_Kfezjquv$y6Dn`?}9;yqFh0lvqHJdsOqOB%Q{Wrvh
z5f$GCw!Tn$NP0c1AjmBZ)St5#LWi3WRVDokTBd50eZqFC`!PG*{`B59rJ)0jat(FV
zxU<2;dk!y2sVn$FX<$(Nj$-^!ve-DNpAH;!Erydh^Kk0Mkz)r0USnDHSP|zPbI6u|
zCE4R;2~Kr>b>-0WSyV8)<CNxPDUN9C9x|gU!J{russ(p{z)2kL@y&?~==fTd`4iWV
zqcD5<y7RaU(=#W8%X-bA)Wb+-j<x`&>6ry^C(A%yEA^1=eG*h<6`HwTt;f-ivsPWn
z9Yhlg_1<md1;iJyZ*zp*SEO;x>IKI{CH{CKRf7H#6I4=uYl+=N3mm7D7M7XzpfDrR
z<C`W5F~9oAgCS=zC{5iY9lNp%(P@f|w2qFUaVx#{wdJG8WbnfFQxO^X{+G3V!|CnF
zc5G7f+^J7!_2gKu*|pC|LLXf;mQKgq6Vq!XueD-LiR-~Zaf68G$cM)17+Sby`94|0
zoe3O6Iry$SRA6D%{+MK@7F@kU_%4@rAwn&cBD+xzlB~+PBwta7XngWYMciMa_(M@I
z`b&v^c!yYBT3d#7Wu!jH3$)=i`7;5U9vSF;kE8re_b=#o?(2r7$~8DC_^8X~Cqw9w
zyG%uM@kjjPR(9=^nP+(BE<@}Y(HwlKYr4|5<{DZP7M?U=+JPhDPU{PrwBQ3CEJ=^7
ztMRT~DsNAed_r}tWsgj`r_j|0=-8=+Diqw9QS&t|4>!I;4S}c$CEs2~&1*A+lH}sD
z^C~itwn5af(8HDZ@$DBMOLjG4D*NCUtPX>y->%8sVNE?|i)QeZc>WBfg|~TfP>TWY
z-I{{BgY~#Iy<1B6W;1e_cGNlgx*A8H3A$`}v>rEfu`4VDR-iT8X^%{-eub`_r~9h+
zAs*RA^h*u<6ywQb;VXoUDzQYiR-yT-clfwF7j?#&X2jcLSgItKi{gXsxmHX(MP?iK
zY;3!qiqBZSxM`x(h9!o^@5~rxW7po*+IMtv@#m!6EF<k=92xC3?2an2M!&g=VQe#c
zVRFRWGA<1{Q$9C6SW6EL)=8=IT(x*t<_Rlhav64F<`3C7SAbIu(xsm2W}v`y{p%M)
zk}&78L$*SbmH0iKwL%(K8NPKQV`DtyAUbr{@>0B0I_`c}NTbWxgc>CW6HS7KP_6BB
zde(|E>?t%frY!IZJ)1-IF3&3Po@UjTUn-igQGH`Cw?!3}zIsh;{Zs}zFrO<bnO%%u
zQ|#KgrQ=Yegyg{(me+`V#wwGV`yO^mUvc$WST7pfLv2#{+W<N>V<bvF=8GF_c_xoe
zT*o>m*%?F3%khA=;>CfWB5ZQVM3U7y1<NnASv`wS!;!`b*3@Bzm`7u*2%Rm#y5!8f
z%bf*C*IQ<$O0OGzzV^DK{6PhF5s2K~ap5caz)LHfoqio{H=GC`)T%&n){G%W`^Hhd
zw*K!@P>VC}N{D@>+(F%DD+;F{(!ep+hwH1>HsKr5YjZB>HsI}UwS0%<xuD{BH{bc|
zt(dgh;Mx4UJY1#NTjgq&iQ?<>X3GwL#Da9wMaKR$_$lXJH*@-CeCJfaD3@k0%4&UC
zVfvyG-`VbAnUP$NA6Y<e?3G%)xou;ZX4)8f#4vGMWEnpcG-|sIwD+OZJwry`Q&qS=
z%~hl$w+%NRu5TWe$-!&owLd&!YedXe7c|??6(Y7bt<vQydr-uL#46r}N~|)?UnIaf
zjP&j&y0=S=pqGY69ZDQ!fRjsa1&ehpl4{_6w>5bXkFHf6JHE35r@df$RXbUN1NsBY
zY|XpT!p>xwiG95&=Z<(5O>{r9@;WWExgZloMZ94#nkvLE^j~cbb*;pQDx)@!`-;Nl
z2$jk{y)GO;dwh4L@Lm)Y@=(z#x)pEHDSQsytyqm=e@~)s3l5$ez8zLuhJ9wN4MYW-
zvB9Cv?qe4^Ft6pOoUZ;xY}MEF#GJYnPxzG|Ici3Nb0N>Zgmr#GhD;Be*v7jsb?{kd
z+7%7BHmIIWTfG&>o!nAgwW<uWp5d*_;VMKq^<ojZtu?6ot{2ZdUmMmvtJ0#r@Cmnf
zD9~un5bI^|hO@S5E6U8hvO`$yIkpnH*J2g>3A;G1uDO+2fmdzOY~C$7ghke!`od?M
zh#guj1k*}7aL(tAb;_%&aZu|9(!$1Wd}-95o4z&&SBh=pIFG7u;Am+&$F3@TmwmN}
zoO3(&I#bY_Pt$;Dvty!v>l(mZ&p)}_phC>Y(;JpK^$ZDm3v@3$E5vTOU(#|<x8e1g
zej3Ie&G<-b1oLX6ejGViaC9+l^!Kw`nMD*~%hfF{bGcpkXl+<~DMu5|G3qIhU>Ly5
zufI!Y`B;GuK0(j9NIf{^Wz@us=pdF)2$N9sdWT)!kMC=Z>ciK>4?1e!9l#Sv-aB<`
zHa@yWddJf*@(`wbcjq662C-txtX_6gKc<t&F6H7W!zS~Rk1iMY<0y{h4fl-)u&TGL
zFn>=yHhyBz^?v6dK3|u!rIxxE-}Bn5oA9g+N9P`M$R6v$<=XQ0>2r#ZcxrP`wR1m)
ztLL8^E$hbA(&~@SQMz%$P2;@1@@3e7`v&bk`5|mf$=KUIGKeP|yKlqbAhs=7b~FA=
zHIAn^R$u4q#KX@&oV_+WfQ{D;R29Vwfg6>(p%;BAX1eR_WX;fvCvU&7h!Y`rRyNmf
z{M3erlj?^LcsAi}XJohZJsiL-f2`V>xw8w;*S=a_a;^oRm(7!G@gBq-#}1oq<Y~b5
zbauxR*yzD@v&eJ51qEoC?K?mJu^cz}*WNOp;)ig~ea5_&y~yc6zzs-Mglmj3TdX!J
z0-H}dqiB9I9*Q_yGu-wBnYET8k8BROQtCY|vR@9=8HI8NU#o(amXy+IOa@7|rhQqx
zO5mue^w=%55UGBlzkOb32+45&X6B<+g_2^Wi#2b(!1?)|Io4lW(U*G@5@$SB;Q+@A
zy?WEv$o$=dYd%AgV7lT=U^SaM^liXQ{$niQVY2*^Nr42^#CqzR%Bg|8e%Ab^Cn-qo
ziR;HIdu34gybFIk$_G9tI~)gy^O5QO00D~&jmY+Xi9uC;8@fqz*+ck^8tmU%m~~I8
zAI)AmX?MV(2NmO!6&u9lz@M2tDu_!OHdY$GA4t@Jn`2hN%~y3G{;R9-z8rB#spN>=
zSXF@)WFpKy7EGY%h(P}mZh2U@J-FCpP#A7M%_-68k%QH~D#w~mh=Ccaf?MbwRgm8w
zvu(rPF(gQOx6tpU2DgOXCl0-kfP;Q@-g2~(5G4>8xqj6I3R+ps-*cD;<|6*E-Lg^w
z))j7jZDYm>Tjl1nT-!+S=PACkf4uo$Pw{#tU%wtSsK93n-!e|!=K!}bu0gdRAz;5-
zaH#xUH5PoePy90)N6Otco5XhM!2W>V<lCZju=j*uE;iK#@{pL-NNXO>C!I#?u9stW
zK9x5M57i;?qD}I?cx`C0jUmfO&Z5fF+Gtmu1@yRy(o*2c0L4YYbr)DQVO!FzQ19QA
z!Io{?33tXU{H5`hyHkt=NUZ4Y37gh}GkQkV-n64=LPfGEe5(w!E4@Db(p?Y+N4Z)r
z&NSl{Ji?O9@6|y&HII$bEeA|@`R<35$w0K+j*Q{ZUUXPY(QqfV9Ar7&ATx}s1C_@1
z<j9W_Q1zlkI3~0npF7c(k&ros0`FLhKQ!TobK;)0<zWJ#`NzN%9hW3H-K9!OtuMuH
zr6CU%K8S)>#Syz}kzz0|yxjNEzB=5g$KDV*#tlg&qc>I%<FDE7a^jY?B+z_Liry2e
z2n@SFxS!LNg{F%!jBlF7;7IEVb`6dz=<AgfhMlwGpg)|Y{_wIqTuzXZ*?+MD$0naq
z%|FctEy?mChk9rr{)v!;lG!J0VB>Sd-&zG~Rf0`-#?GP{Yg*mouM~i<BqTjnVHEND
zWyL-X<%afkY+B9P{J?TTq~0r(8CGV{O!u?Rp*8byJpsj~*xY#gh>uqT4$NunuGy>#
zfhyC+Eo+saI#%YjLAVOMj!^!TB+CcOPIi9m8y`b122?9K9J%2_M}VNC<RHotI6GAA
zs0#7nxvHMN<v4WsH%lkt_QpuI2K!T%1I2AFJnAkVcpewj(rFif-SvkQ<S%BU`vxvE
zrME{=@0lU3-{WTy%iNx+;J1~SfrsY<*CTGoz4@R*XNv$dTt8*!-tY+-`&bOVy(SF@
zSdTpui(C$AF^*y<%$I{*SMXj5&yRSG*N6Q1Yl6UCa{~!p5`&m|e7x*>DW*L?qAZ~(
z1sRW@fU`LttcWVYs=LHN(YM&o-!2ucYIY2XCr*8a^1HU*mJx$C_shM8>7qbI3fz5O
zG80Qpm<;dr5eLr-zbwmmX6V;4p|2lpz(LHi{%JcnL7qOFt-zTJ%1ifO>MoE0Nh;Pt
zpH@kDu5^*A@gy^x+pFLGdx|LZlpeq7Gs+EZslBhWy~N<|`CasFzIoVXce3G|vtq<5
z_t@9UG>x>1w#JPTr~lQ~pPnzQE5mCxkyX=(TcFEBca*C>Q$VLr$ldS}HK>Ve1%{sC
zf(3m4wB8*a*y&vt$IwO%HRoOw1sBu8lhpmw^-5I0v@eW$aC9Dtk6Nu?F~tBHVV?Gh
zG5oM$|9)4{-~;ZdmwVQq<A*&>+Ygf6`Vr5J*~I-%+z_!wvNo254&LzZ$PMhJg^YU)
zA0KQIfg@V>_R0(Nur|2d`ovdyIDa-{`p{((_}jWw=ywW$-jpBJJeMd;ik7^@E?l6W
zncj9lwgSsl*kArcCk{NhHW^QazoLBxG7XpHmw__pqcGFx86?#Bs77T*1o97m9jbMZ
zfY-&<=pNML0J-Ju!HyMJn$yutvq=;b?>=_qV6MP%i7|KBvq-SlygR%vnF7jRI41|f
zxnVnB@?fw&52)RwdKi6+1E%?Yzp&j}1X6gjchMZ-hQpC|W|2eOP%CxVVtu#}R4diC
zS;nx!Y@31ShnP$h`b_tWTTmw|xx*2nMBJ`uA2!yS8y-g?W2rIA_lbaKa>D*+c~r2)
z_suxvHZv$q@?Fho7J;}m*J+%)dBNKGxUIRw6e>HBy?N`1C`k3XbH-SULEm!6bhA5D
zaI*gB5ldQ8@Y}i9_|OL8*Cy-L^xp-Z;)P2(tig%=5O|4p^CseH`bP+xR?$`BwkhFZ
zmg|0LDAG`y@=>D!>jQO+S3}#-z&5Fo@mx``c4EkRL*fLJ`XUA2W6NP}K!W<HbRIhR
zvBcdiHXH9X?vBY+6@!f~O&3Kr7a*6;#;!fmY53V}WVUDMDB5NI=+w=GEL=S?*kx4A
z4$I!UPF^S}z~_!E2rfI8f;;zFWn}O$!3UnaJ*P6nVVl_Hgw;_6_&j&o4%$}(=-~La
zc4d+<h&=a=>AuVasmCoh>R3uc{N_t+(ksQmqiA<rab*sY-}a_XX_+`oi>4i<tD(S7
z19aM2R05hrR#g^S67A9>gXpm^6xmof(RlL$t&Pd<A__lvzBxKUPJDwNHgPVTOpphi
zcbU!1AA~@-<l=C+?Q-afwotNP5P%W?lZ9jF3h)cV#7F)S@6coz=^fQqM_i#eFMpSx
z18O{`^+iz@VqvzYZ!xdNWLnS6dyz!^T=$yR@^zpSHx8GtFA)Trs2xwfcD_Te+IeK6
zzNR4;L8aL};q-9!^|6z)$2g&MUDF=x1QI--k7DzA#|o(MDc`H}RFFxgxvkL5K-|);
zf1&TtfGn84st!reK-$F*Z`-%f0y$B@Ub68uK45EBw<3=N)-ycrSQpTTy4v|($1$)#
ztWyYQ^!z0H${N+VHl6`g_Va3`CDX$SRe`M=7picpzQ&WJ)l;bI_mvho(E@OvZ~s9p
zV=fR?S;KiyNdUwsL63J;_n>vPx1W^c@<5Ve+^B6Y8`uibFl9fYhm-d=M=;tvLPn*l
zc#0)Cz|QE++O_f&aA(@JZT0~>Y#tPJBP&lKQ~HzRGa+(N_j^{+oeoYoBdZzY)>Vqy
zUTH71`*A{DxPs03cH&mWUz(Omiy3@B=G~B!C4ub6{&3qgdNAaF=RV)nfc#ok3BDa;
zhm?FY!u`G)1wM}pwSV4<AA3BDdw-J;CfK5~H=n16q734~Fsu~&D{kr3d0d8$n5X%f
zF0ez$Hu~$OS?q9#?NRQ5c~-DEpsb>tUXEMJxrC}`cwwi<vx%TvT+n*6R$XehIG8uA
z_1aZK{A(btpubX}27m1B3JCFQLDh|#+?3zh!JsZ8VVP$lT6L@FNYa{FG;VnDp%Lpi
zqS_H2y5blum?oxd*zI^5A6OS8&TGR2#|)c}`hF<JY7Zr{B!)hqLtXv3IHn3oKe|UB
zq{t4zRn#`M!)<8!j&(6z4<F!6rnRSTQRku*$_Wp<!W3j0tH10VHycdq<+i?H&&HHj
zJ@#iSC(%SDtw{0fCX8&Ell1RU!+yECpozkK%q~^0`YLrA1)FkRtS-NfS$OqXQf)Hu
z;oUb?rSoWDouK*Zxof$2u1HO(`6DZIW*@Ph>SF@l%17S3O(Upa!s_vx+6J7}y^K$O
z${*op8C#>B$Ix4<UWxu0M!0`O!!(wc1j3&Nu3uWi2;@tJQJSu`SiLD}nks7^y)9_E
z?rYkPCL_i#pW=Uml=Nng3{mIftFq6$j<ZOB_X*+IkQ!#FK6gfccjO$}HY2w3<zyM&
zbN48N<Y)=L`8vJvjdm?wo87Pc*|i#nT{kFJteit_b4)E)<e1<R``cQ9`SUofdgMq;
zR58vJJgG4HvIUPie0;zyDFsb)G7&edyHSN=z`{*(I*vRRoH-XeiBz6aH}>7?MmAh4
zUUW~B;0Ed9-YrxkXnN=ALu<>a!R&F}j>xUKcteGpBHMv^v_X(Hp)X(##fXGQ*b={^
zJUX2ipHhE8r$ypk3tXH+MHW}J=N~R0*SV%uPW5CEzqMe+1k@nbU+tJa#0D>~Y;Bgy
zXNB$y(Jy4Kvx6I@d+fbnIle0<Hh)s+Ga5LdWVb~^34CkcZr=ZF7)1?R^_}{S_?Lm9
ziDjX&4%;r=n<x+~!(|3m#ll?Uh&9LQf{)o8(!Zv`>(ZEt!uBQ$iO^3X%luJupKZ;U
z=e@vwbq{)|zbm|H58DVj-5yZvAe>G-_s=?Ijx-@_ZEc_9(oJ|wN49#N<rC)ZUoUBR
zrv;^y-@6biH;YR4jfuQbA4lq;(%G-e+py5s*e08vMr<Egte0F@hI`%kBCpZ3;Hroj
z2c_0ac$jyOxwzX1;+H$ur>K&Lx2=><kgsQj7aAe;H5#*s?e6HA0hbDts-|yI9bALm
zzaFbhx>kdRUjziX9hgMBiY%k)eA01ekd%nviyrj5Qas(U6HS==T)Y3N)h+nVy3U~4
z17DEYld0H?-9<QNa-jK1au?D*pD6Q+_+8q>!>xQtpaQRNo}Vs~>%ekS5_y-|C()4z
zt5NEU#HrLG&ZM776^!@HmZzMpMRD_<mQ{B3SmD_&ZR;Zo=nrfA&D9>&c;{aC>`|%=
zB-czkODdg4yZn#qp}`w?|IzEkVIfoKz}fMxE59`$weveJIlF&CX;s_s33o*ZYFuYN
zQ8<JW7+*=VUl>M#rHN^gq*lxrFZ`sxs{;*_ev=HyoJOYIy60{OH{+&>Ml0nFJt*5&
zm~qFtEEM1#ViD3=gQ^bryp=rNg;#7!AMHIN2epocO{yWY$lX@A{jPlzmNjmpw#jP4
zBga`{GWDwQ^2nLv&zNS=r(?b@q3frSLZy*I4`l{1Y!g?QS>A{@QC=$zo44Yzy53Wb
zZnTiYD9I!$!3o~=ZQ;|9j?D5~yu%NDMU~s_H+Oaw;P|zBsFfKCu@9}r#$_B0xM4!0
z`!lgmgq+}b_(`h_TTVXMTNc}l{Yp<39Oj%v^BrBStXrG0`(O(;Sw#&i&DoDel8Mu5
zeVp=50vCU{xXe6K5PHskR7w%(MN~8+qA!&d;dIFFJba!iU>_QwU$~<VGwg~!Dxpe(
zKJli~4kliZZF8BQTd4w72RVzH1T;Wfp%$_1e}|3q(@~MUDolqQzqPuu7CY~zoLo`A
zfTB75A8Xdu;ea>H3C@gF*s3!-P3u7=9&iYiC=$_tivwf{*(>E(c5LhA{jw5}(x>WV
zR;B@`nm=B9*wKT=IB#tZkfA`LdtoBodIgY)V^J&fslpw}N4@ATszc1><N^xj2OFHI
zcPv&JNOcztX{^g|WmrcU-|7auO_}|3N|Gv6@J#EfL{#HHcAZwZ|0*BnbTAc_PY8o}
zrgClZEn&ErdAEOzI3F^zv~-tesKL(X+6yE8BEa*yuTkT4E^eEWcQ9G;9EXT=P!hS7
zfLZiS%O?&BgsyEYWLhr^(Sa0Sn+rlPoggIcd`trpqq+{htsu_RH$R6+W~o73-Sg-5
zTEcMW)|&Owxx8@p>`G00gB}#4sk=WVoD-O-r^NQM5&s_Qtx8#ywSe9S?pC8$>_Ge+
z)ckx5OK;`<lB7`c^N>j2s~6vWTl^5!OUYlVsnq!SsMLNbp!Oq5X#Di_eko&Ex^^OL
zEd4|OYxUCnC#si@{y(aPc#J9fL{~`sLOd_2{!HGc{xjJ?O=(*G1MyCj5{7=HbL@Yl
z2}@z7%zi!&zV7yZj(eOv9VkpB3R#N8$L`_m;=j><pVJ`=Z^wOpW=`LWIegt*_xz=t
znSe7n?(y~Z^!D@ibe1An2oas4FiTPMG5b5X+B*2UIe7V-IsK^6BG7*R4*qU_{%($b
z#LJHqCIZQ{=+EB6+tt|Qd%?1Ao_>_@^w9GQ@q0)c3sK1YQVeANzxR(qO@Mejc>JY}
zr5S{NGWRpZLz4ZI&!43vensLHOEQI;MBLjfDiY77TT~Qop7K~K%1V;-OX9W1A3py+
z`0q}C=kX<A(nkreOS9l-YSHW8OaH;^L&7T$&xM=cyxyi3qA-6S=`W%E?#_kBOZqmo
z^bnO0F}sK_;o+P92b*c%Y!(wX#Yxf)-)w4;80h%_Co$j6#vAf9Ehj(*+9cV<Im$=O
zR$^)qQ6YVcN+n^7k7t;$B}<ZiCn`i?;{Rq#_7^sPVez}slZYn_5om*-lbi3)8NV2s
zr78cj^jDt)-+W$;`xcWN`+xNL>qPn0X^uVNbZN=(F)Xd4MW=r+{ncsU?Ztsse{;&y
zmr6y<D`^IvkFiwT#8Ok_<MZ})Rd(=kAeM{rx0R#3-`UsC&D)EQRzq1sS(C&k_cw&!
zUJqqrk?q~*?62(WeL&gM+26s*!QX-J2Kk7%{~;e|aeZ+=f8TwM{^E+_E^f{qPJZJ0
z+m48PIe0Es{dFx8pdXDqyuDnBoYy`N4~N|z1lGmD!_Qe!e93{h{*izH#Xk-GZT0?t
zTl%@)f0`gh;P(I2onLo{|1?8{=HCbPb2ItV#{V{`-?x`Pn~6Jl?^}%GuL1o}*T3vh
z|F1@N1pMmdpJ>|=K;m2XW7a5n`ilOTW}<4!YU;}BqO!a9xp_E=>gYIYs5<JZ>S<``
yYCAh>sXJ;ox;SV%YH04(ady$rP*>AXS9jFYl;igxw&lNkNmO(s(t6^R^#239QVC4}

diff --git a/tests/gentropy/method/test_colocalisation_method.py b/tests/gentropy/method/test_colocalisation_method.py
index 1f52244be..d311d88ad 100644
--- a/tests/gentropy/method/test_colocalisation_method.py
+++ b/tests/gentropy/method/test_colocalisation_method.py
@@ -4,10 +4,12 @@
 
 from typing import Any
 
+import pytest
 from gentropy.dataset.colocalisation import Colocalisation
 from gentropy.dataset.study_locus_overlap import StudyLocusOverlap
 from gentropy.method.colocalisation import Coloc, ECaviar
-from pyspark.sql import functions as f
+from pandas.testing import assert_frame_equal
+from pyspark.sql import SparkSession
 
 
 def test_coloc(mock_study_locus_overlap: StudyLocusOverlap) -> None:
@@ -15,43 +17,92 @@ def test_coloc(mock_study_locus_overlap: StudyLocusOverlap) -> None:
     assert isinstance(Coloc.colocalise(mock_study_locus_overlap), Colocalisation)
 
 
-def test_coloc_colocalise(
-    sample_data_for_coloc: list[Any],
-    threshold: float = 1e-4,
-) -> None:
-    """Compare COLOC results with R implementation, using provided sample dataset from R package (StudyLocusOverlap)."""
-    test_overlap_df = sample_data_for_coloc[0]
-    test_overlap = StudyLocusOverlap(
-        _df=test_overlap_df, _schema=StudyLocusOverlap.get_schema()
-    )
-    test_result = Coloc.colocalise(test_overlap)
-    expected = sample_data_for_coloc[1]
-    difference = test_result.df.select("h0", "h1", "h2", "h3", "h4").subtract(expected)
-    for col in difference.columns:
-        assert difference.filter(f.abs(f.col(col)) > threshold).count() == 0
-
-
-def test_single_snp_coloc(
-    sample_data_for_coloc: list[Any],
+@pytest.mark.parametrize(
+    "observed_data, expected_data",
+    [
+        # associations with a single overlapping SNP
+        (
+            # observed overlap
+            [
+                {
+                    "leftStudyLocusId": 1,
+                    "rightStudyLocusId": 2,
+                    "chromosome": "1",
+                    "tagVariantId": "snp",
+                    "statistics": {"left_logBF": 10.3, "right_logBF": 10.5},
+                },
+            ],
+            # expected coloc
+            [
+                {
+                    "h0": 9.254841951638903e-5,
+                    "h1": 2.7517068829182966e-4,
+                    "h2": 3.3609423764447284e-4,
+                    "h3": 9.254841952564387e-13,
+                    "h4": 0.9992961866536217,
+                },
+            ],
+        ),
+        # associations with multiple overlapping SNPs
+        (
+            # observed overlap
+            [
+                {
+                    "leftStudyLocusId": 1,
+                    "rightStudyLocusId": 2,
+                    "chromosome": "1",
+                    "tagVariantId": "snp1",
+                    "statistics": {"left_logBF": 10.3, "right_logBF": 10.5},
+                },
+                {
+                    "leftStudyLocusId": 1,
+                    "rightStudyLocusId": 2,
+                    "chromosome": "1",
+                    "tagVariantId": "snp2",
+                    "statistics": {"left_logBF": 10.3, "right_logBF": 10.5},
+                },
+            ],
+            # expected coloc
+            [
+                {
+                    "h0": 4.6230151407950416e-5,
+                    "h1": 2.749086942648107e-4,
+                    "h2": 3.357742374172504e-4,
+                    "h3": 9.983447421747411e-4,
+                    "h4": 0.9983447421747356,
+                },
+            ],
+        ),
+    ],
+)
+def test_coloc_semantic(
+    spark: SparkSession,
+    observed_data: list[Any],
+    expected_data: list[Any],
     threshold: float = 1e-5,
 ) -> None:
-    """Test edge case of coloc where only one causal SNP is present in the StudyLocusOverlap."""
-    test_overlap_df = sample_data_for_coloc[2]
-    test_overlap = StudyLocusOverlap(
-        _df=test_overlap_df.select(
-            "leftStudyLocusId",
-            "rightStudyLocusId",
-            "chromosome",
-            "tagVariantId",
-            f.struct(f.col("left_logBF"), f.col("right_logBF")).alias("statistics"),
-        ),
+    """Test our COLOC with the implementation in R."""
+    observed_overlap = StudyLocusOverlap(
+        _df=spark.createDataFrame(observed_data, schema=StudyLocusOverlap.get_schema()),
         _schema=StudyLocusOverlap.get_schema(),
     )
-    test_result = Coloc.colocalise(test_overlap)
-    expected = sample_data_for_coloc[3]
-    difference = test_result.df.select("h0", "h1", "h2", "h3", "h4").subtract(expected)
-    for col in difference.columns:
-        assert difference.filter(f.abs(f.col(col)) > threshold).count() == 0
+    observed_coloc_pdf = (
+        Coloc.colocalise(observed_overlap)
+        .df.select("h0", "h1", "h2", "h3", "h4")
+        .toPandas()
+    )
+    expected_coloc_pdf = (
+        spark.createDataFrame(expected_data)
+        .select("h0", "h1", "h2", "h3", "h4")
+        .toPandas()
+    )
+
+    assert_frame_equal(
+        observed_coloc_pdf,
+        expected_coloc_pdf,
+        check_exact=False,
+        check_dtype=True,
+    )
 
 
 def test_ecaviar(mock_study_locus_overlap: StudyLocusOverlap) -> None: