From 2e227e4960d022129b190cb3b1ba99aa6da4a33b Mon Sep 17 00:00:00 2001
From: emmarousseau <emmarou1@icloud.com>
Date: Mon, 13 May 2024 09:18:30 +0200
Subject: [PATCH] Update test 1 and test data, fix some arg types in config and
 script

---
 src/umi_tools/umi_tools_dedup/config.vsh.yaml |   2 +-
 src/umi_tools/umi_tools_dedup/help.txt        |   8 +++---
 src/umi_tools/umi_tools_dedup/script.sh       |   4 +--
 src/umi_tools/umi_tools_dedup/test.sh         |  24 +++++++++++++++---
 .../test_data/dedup_edit_distance.tsv         |   5 ++++
 .../test_data/dedup_per_umi.tsv               |   6 +++++
 .../test_data/dedup_per_umi_per_position.tsv  |   5 ++++
 .../umi_tools_dedup/test_data/deduped.bam     | Bin 2154 -> 840 bytes
 .../umi_tools_dedup/test_data/sample.bam      | Bin 3268 -> 1584 bytes
 .../umi_tools_dedup/test_data/sample.bam.bai  | Bin 889032 -> 2656 bytes
 .../umi_tools_dedup/test_data/script.sh       |   5 ++--
 11 files changed, 46 insertions(+), 13 deletions(-)
 create mode 100644 src/umi_tools/umi_tools_dedup/test_data/dedup_edit_distance.tsv
 create mode 100644 src/umi_tools/umi_tools_dedup/test_data/dedup_per_umi.tsv
 create mode 100644 src/umi_tools/umi_tools_dedup/test_data/dedup_per_umi_per_position.tsv

diff --git a/src/umi_tools/umi_tools_dedup/config.vsh.yaml b/src/umi_tools/umi_tools_dedup/config.vsh.yaml
index 312d5078..0614cbb2 100644
--- a/src/umi_tools/umi_tools_dedup/config.vsh.yaml
+++ b/src/umi_tools/umi_tools_dedup/config.vsh.yaml
@@ -134,7 +134,7 @@ argument_groups:
           tags are “NH”, “X0” and “XT”. If not specified, the read with the highest
           mapping quality will be selected.
       - name: --read_length
-        type: integer
+        type: boolean_true
         description: |
           Use the read length as a criteria when deduping, for e.g sRNA-Seq.
   
diff --git a/src/umi_tools/umi_tools_dedup/help.txt b/src/umi_tools/umi_tools_dedup/help.txt
index acbab88e..87baf322 100644
--- a/src/umi_tools/umi_tools_dedup/help.txt
+++ b/src/umi_tools/umi_tools_dedup/help.txt
@@ -33,15 +33,13 @@ Dedup Options:
       --output-stats=<prefix>             One can use the edit distance between UMIs at the same position as an quality control for the 
                                           deduplication process by comparing with a null expectation of random sampling. For the random
                                           sampling, the observed frequency of UMIs is used to more reasonably model the null expectation.
-
+                                          Use this option to generate a stats outfiles called: 
+                                                [PREFIX]_stats_edit_distance.tsv   
+                                                      Reports the (binned) average edit distance between the UMIs at each position.
                                           In addition, this option will trigger reporting of further summary statistics for the UMIs which
                                           may be informative for selecting the optimal deduplication method or debugging.
                                           Each unique UMI sequence may be observed [0-many] times at multiple positions in the BAM. The
                                           following files report the distribution for the frequencies of each UMI.
-
-                                          Use this option to generate a stats outfiles called: 
-                                                [PREFIX]_stats_edit_distance.tsv   
-                                                      Reports the (binned) average edit distance between the UMIs at each position.
                                                 [PREFIX]_stats_per_umi_per_position.tsv
                                                       Tabulates the counts for unique combinations of UMI and position.
                                                 [PREFIX]_stats_per_umi_per.tsv
diff --git a/src/umi_tools/umi_tools_dedup/script.sh b/src/umi_tools/umi_tools_dedup/script.sh
index 00889600..8aa89d10 100644
--- a/src/umi_tools/umi_tools_dedup/script.sh
+++ b/src/umi_tools/umi_tools_dedup/script.sh
@@ -20,7 +20,7 @@ test_dir="${metal_executable}/test_data"
 [[ "$par_subset" == "false" ]] && unset par_subset
 [[ "$par_log2stderr" == "false" ]] && unset par_log2stderr
 [[ "$par_get_output_stats" == "false" ]] && unset par_get_output_stats
-
+[[ "$par_read_length" == "false" ]] && unset par_read_length
 
 umi_tools dedup \
     --stdin "$par_input" \
@@ -43,7 +43,7 @@ umi_tools dedup \
     ${par_spliced_is_unique:+--spliced-is-unique} \
     ${par_soft_clip_threshold:+--soft-clip-threshold "$par_soft_clip_threshold"} \
     ${par_multimapping_detection_method:+--multimapping-detection-method "$par_multimapping_detection_method"} \
-    ${par_read_length:+--read-length "$par_read_length"} \
+    ${par_read_length:+--read-length} \
     ${par_per_gene:+--per-gene} \
     ${par_gene_tag:+--gene-tag "$par_gene_tag"} \
     ${par_assigned_status_tag:+--assigned-status-tag "$par_assigned_status_tag"} \
diff --git a/src/umi_tools/umi_tools_dedup/test.sh b/src/umi_tools/umi_tools_dedup/test.sh
index 4b83ff5e..1b5a9053 100644
--- a/src/umi_tools/umi_tools_dedup/test.sh
+++ b/src/umi_tools/umi_tools_dedup/test.sh
@@ -7,27 +7,45 @@ mkdir -p "$out_dir"
 
 ############################################################################################
 
-echo ">>> Test 1: Basic usage of $meta_functionality_name"
+echo ">>> Test 1: Basic usage of $meta_functionality_name with statistics output"
 
 "$meta_executable" \
   --paired \
   --input "$test_dir/sample.bam" \
   --bai "$test_dir/sample.bam.bai" \
-  --output "$out_dir/deduped.bam"
+  --output "$out_dir/deduped.bam" \
+  --output_stats deduped \
+  --random_seed 1
 
 echo ">>> Checking whether output exists"
 [ ! -f "$out_dir/deduped.bam" ] && echo "File 'deduped.bam' does not exist!" && exit 1
+[ ! -f "$out_dir/deduped_edit_distance.tsv" ] && echo "File 'deduped_edit_distance.tsv' does not exist!" && exit 1
 
 echo ">>> Checking whether output is non-empty"
 [ ! -s "$out_dir/deduped.bam" ] && echo "File 'deduped.bam' is empty!" && exit 1
+[ ! -s "$out_dir/deduped_edit_distance.tsv" ] && echo "File 'deduped_edit_distance.tsv' is empty!" && exit 1
 
 echo ">>> Checking whether output is correct"
 diff "$out_dir/deduped.bam" "$test_dir/deduped.bam" || \
     (echo "Output file deduped.bam does not match expected output" && exit 1)
-
+diff "$out_dir/deduped_edit_distance.tsv" "$test_dir/deduped_edit_distance.tsv" || \
+    (echo "Output file deduped_edit_distance.tsv does not match expected output" && exit 1)
 
 ############################################################################################
 
+echo ">>> Test 2: $meta_functionality_name"
+
+"$meta_executable" \
+  --paired \
+  --input "$test_dir/sample.bam" \
+  --bai "$test_dir/sample.bam.bai" \
+  --output "$out_dir/deduped.bam" \
+  --random_seed 1 \
+
+
+echo ">>> Checking whether output exists"
+[ ! -f "$out_dir/deduped.bam" ] && echo "File 'deduped.bam' does not exist!" && exit 1
+[]
 rm -rf "$out_dir"
 
 echo "All tests succeeded!"
diff --git a/src/umi_tools/umi_tools_dedup/test_data/dedup_edit_distance.tsv b/src/umi_tools/umi_tools_dedup/test_data/dedup_edit_distance.tsv
new file mode 100644
index 00000000..89684b04
--- /dev/null
+++ b/src/umi_tools/umi_tools_dedup/test_data/dedup_edit_distance.tsv
@@ -0,0 +1,5 @@
+unique	unique_null	directional	directional_null	edit_distance
+3	3	4	4	Single_UMI
+0	1	0	0	0
+1	0	0	0	1
+0	0	0	0	2
diff --git a/src/umi_tools/umi_tools_dedup/test_data/dedup_per_umi.tsv b/src/umi_tools/umi_tools_dedup/test_data/dedup_per_umi.tsv
new file mode 100644
index 00000000..a1d364e2
--- /dev/null
+++ b/src/umi_tools/umi_tools_dedup/test_data/dedup_per_umi.tsv
@@ -0,0 +1,6 @@
+UMI	median_counts_pre	times_observed_pre	total_counts_pre	median_counts_post	times_observed_post	total_counts_post
+ACCGGTTTA	74	1	74	74	1	74
+ACTGGTTTC	48	1	48	49	1	49
+AGCGGTTAC	1	1	1	1	1	1
+CCAGGTTCT	1	1	1	1	1	1
+TCTGGTTTC	1	1	1	0	0	0
diff --git a/src/umi_tools/umi_tools_dedup/test_data/dedup_per_umi_per_position.tsv b/src/umi_tools/umi_tools_dedup/test_data/dedup_per_umi_per_position.tsv
new file mode 100644
index 00000000..d9211d0a
--- /dev/null
+++ b/src/umi_tools/umi_tools_dedup/test_data/dedup_per_umi_per_position.tsv
@@ -0,0 +1,5 @@
+counts	instances_pre	instances_post
+1	3	2
+48	1	0
+49	0	1
+74	1	1
diff --git a/src/umi_tools/umi_tools_dedup/test_data/deduped.bam b/src/umi_tools/umi_tools_dedup/test_data/deduped.bam
index 5ffd9fc26f6dfaea2dd1f4cfe2c1cb18d93c59ba..a82e6c81a47c472acd45c379a42b294f31966e87 100644
GIT binary patch
delta 825
zcmV-91IGO75Xc69ABzYC000000RIL6LPG)oTmpTPxo;Ce6o(%~N!ZX)Ks!nUQdsZY
z$0FqxhZF~cB}fF#I^M)9vRP|w+4M9BQBd+9BvM3=D1e@t1|cL$3MeR`WxbciDJU$>
z_dd^?_nX=EwdT3UGXSf%H=PIDK51a5eb?VhlC&Eg1;cQEVYPkFX>a>`y_7gx*&mD|
zMsP-LfpbHRF^mdfn_>?onlmk=ZD?tzP=aCGu)<)PZ6Z_W@WKdUxR%^DA~%?E!;P9U
za$<~T#1JucjLL<eib_Fjp>u(iqL@tyGZB(7PWjZtMHEdG*H|k{6iI{;#wg7!Q3Me}
zGMojqM3F;(P{9-x#u9l9!i^vbGy4ox?Vwx;`;0sWVMH;lm7T*Bx5%-QMA`S-%!SO}
z(JJl~MJWZQ#(u%wG91pHJbUxjE$7arzn+YTQCNIQq(PmHEq}3-*O2Ru)1coEQ`89t
z(VmO5%CMJ)L3e;Saz}_FH%25YKS2a}(f+`Tqclx_Qtu!<O5$+f6^*<&HpuISBX7U!
zNtSgP1o2`uK(-zx$>G3hZCmTs^kAl#tnz3Sg=6IQQ1~>6`-fqp6U1lxHOSeH_4<E3
zJ8k4HTn2#I>^J#_;N;K8i*qGlxccJ5d_@Fa;f<Oo_{)Ax3|@DaYU0qpKdMQ<vRIm{
zNJ8U(#@E@36nuLBrX~$5m!g^sd^>+KQw~O9y!Lsa1O~@j&zurCTzhx30s*JbuT&u6
zXZqtp2@0HtwbAhR`j`0<47^+oD`>*$N_`tU@b>$!YR@h_I{pp+0PvNx?2rTi03VA8
z1ONa4009360763o0L}n~&^-=<KoAAs0WV;GVy!pO><`P2S~!jgHdwGz7%ez}=g`8d
z7!Tkryb&%7A)%31%`{(Lo`p7_<r?O-oCPrTf8C|IZCNT|5|f#g5k`qPu8=6tnHfP$
z0M1BZakxZ^6&pAe6mrMjpIcben=w{u$0kNrMA*2*unT%uDijoZ*LLppy?1r3bK##b
zn`S~rT2+|)x<~_-Bb0i;K3+%1-o{v|q|j07R$#vZ!8hz#K?&-%?GJ$Vim+V)lS~L6
DG)a52

literal 2154
zcmV-w2$lCAiwFb&00000{{{d;LjnL-0#%Z`YZO5o$A21<xMydf?T|(>?(MuEE98w2
zNX}?L6RhrX8?%_r9=pe#y^V;4rT;+!!CoxX-r7b)u(Z&^LR;td#oWTee14yuncsJ2
z*H*XAU7rD1xwGLt?1ZF^y`6jEewt?e__#ZaS}QyEy`4_Df0z+((|%wSF@iJd3S1a!
zjA2v=*OYT8(VS@^T|+BFg%S+AhLr|WJBdu~!%HKG;aYOnh{9mT4L54a$cZtU5kth(
zHmVSUDk=qWg)RhEiefe;%tT1SIOS6vmr*oPTw|>qQ6>>a7^5_EL>WW~$*>LTh%$$u
zf+;GDBZ?S=8$lFi?jEY1LAem_9z_hoh+<kRH-{-tkz*x^a?f|G5YpbGQ`{|!QVL9s
zdxLvbIBc)n-ud=z@9svpmX3#URQ^e%O}+KaaG_V!knfMP?qCpQsMpQoeIMJ(@Gy(I
z{Ty-Rj}S$Ef`~0YMFa)$K^`P=mZe$nBsxx$C=beof+R60=traApdUzPSIN7{LfwEJ
z57YE0_qIFEcze1rQ+8H+G>W4!@_Wcnk*y7mqIR#FAU{XAjqM{)^sFhjE(5@<eJ6he
zoc{fEajpUkS6_aduZh5GyxtIna4Bqv!JGbKLmUPVMhyvA5{q*+Noe2vHd~W|&mZ14
zq+$6|+>n9q=TB#<#wd(ezqBe~aI*Q_tAN9`_qS>gaQ5O#4HAB3KQC0Ez<bmf4gYR@
zov*;atCgsREjU|lPGc9|{rFwa*@MR?f8akV_XD<G1ONaZiwFb&00000{{{d;LjnNx
z1(lauh*eb>$JaSW8+FRm$x&*~@mP85GOYW>u-a{5r;Ik2U3rPNm{11EfG;LR`PNej
zB?k2{PZ3cmK}kepDfOTa`cnB21`-(sNj*fRv-{S*tg}y^Y1x4_oOw7uzV%)H-+w)f
z^3R3GC+D`(dAf!cWdEnd$9|s}AE#I-VZ<;Yl5;9e631Z_#c>d#sRlw@g0Q}6{Wcjg
z8552+Zwl*U@WUFM1?Q(`HPh3#d%*BlNFkCS=f)vASj)II;8GS63#~m=K*xUj0-*5<
z57ZDYj1ox{l~6n@gCZNOvPd)vKdHkx|H2cfmF}<m62KS}Lh%HyUL@dZ=2*de?8ig(
zW;!)gE0R0|6Rs{eW^nh&BEz^u+Pe!ADtC7a{b>)I-Ao5pd7xS;OgSzK6~#T`ERl!T
z9s}fxH6EnIj9>+E=IIP0G+B2@X1b8*!R^;I(~s`-RftkUm`cJZ&Z3Oar4EvJqJ$np
z=e|Rb&5@gZ*^~qmDiV8dTc8MyHXKqgS9FU!^Fh1ON>8;sNJ1IM+L$B?a^_6QoMjm=
zt%lzG`Bo2;YQu@tNfZWIZ^r2R%FShloG6ed21fw->}?(-1Zk+!NgS6qZ`a&x5*&yq
zo1%Id=``C6H9Fv{5ut#^2HG==GzidS%|%L?yZmf84?ME9)=b}D;%iQrAwp;pgjw0c
z03EE<NOd9ifB1Bxnf|`a6Q$x>NeulSl|wXIY0hrso|CTu^3Ap{N-njL3_2;Cu|%T<
z@{M2Lg-*KM_T>^wB{U--A9=(7S2A}eUqYBWmwT9!2}%fa^@+3Umd%-|F&c-T^uhW1
zZ`2q{M_RtI#2j0V%ZX4P+!1*WDj#9jNS1o&opaEiM|^UuIhLC8B+B~}?tZCLb=jFr
zi=037>FidzZjHA$ji82V$Bbr}%sFuShjxO1oV~?2nT#`TI5cNaY>ge2=Hy+-4b5vq
zQUzM9Ksn=LGLKd|iFF}=xi%!@oMU49J&!W#F+yVna-e-pNJ+S{xeUWn-z$4a#uB?d
zG#AdP={7)5`{Y<dg@lr2HltqqvP4eoVt^d<#Yu>f)XrtD#?YRVm8**`<hI4TV1~{u
z^C2;2jHoi3P@prFu{185^WZ#l&%J0U4F-IIL<Lv{(D%68C)i{%9Z;QRbG(2qe0?tn
z=aQvfn<OPO8X76+b~Kqy5+U#Sq<%P`9k_y;>3xHq+w;tOfx_o2GY)2`kwu=p`>wub
z`sc7OP@}mr6s{i@qI7>nj%62e_x8`{HPg-Oyw;0rL9ADZijKcJN}JB>;k<gWAIiSZ
zr*jI{I=BTWds(Ey(PZf)A<3=RXhB6+A`qpgeAWxxhk$;vcA?zu$1B~G6&}R}bN9d}
z;Jt*;dvQ#al6DEl-3}y^WExTyVCTCt?iS0HJ^|&@7{+Pomf|k&sUl%q0&akF;`!%l
zt#s{De*&1!_0A1_WkQLozGXJPJq$zt_K>fkG|^I8-yIaoqj&W!@|)xjFg7cPeG+O6
zv5T{O%S33Z^X4ElB+L!dd^mr7oWO{FwZs!B)m#wpilvmqJ^bfbfqZo3$vMq*tMCMB
zFi|j0PVitObppjP?F{{mR-cwh82aS{9w_WhiET$Y!*sl|leml1B3}s+G}FeFKBUG<
zYguX@lzp=Dh@%niigWSm_drRmVBc1fFr;AYidP4W{n1K3jVruGp1j<Ed~TsWdjsy%
z?xe!s{7Xe;z&n9jj=azhf$sB3o?!+H%-vfi=u!CpiR6??BUlpU(+LTEv4WKD?iO-v
zhk?5fGhaFlWf<Ii@$P`Kw>y$YOprg{v%4>XvvC^1D2?3VtGpC6bKiO8VS^m4C|>8K
zZIOL@7xuN%mwk5!Ye$)(rnGjzoT@Y^b@?2y4GKH7nQmC*dmB>43<Ikil#k-8d)^{%
zf;0DrC!yyruJV~$!6Bcx%xBPRD~%S&w@y#P{#frzCIR0-G^K7ddwHgR0Md6qGr17}
g03VA81ONa4009360763o02=@U00000000000FmJWb^rhX

diff --git a/src/umi_tools/umi_tools_dedup/test_data/sample.bam b/src/umi_tools/umi_tools_dedup/test_data/sample.bam
index 929d93b0a7b427ea9072b9f4a814363ecfc0f8d9..32192fc6e9bb58449cc20acd27d747a6c60399d5 100644
GIT binary patch
literal 1584
zcmV-02G98)iwFb&00000{{{d;LjnL<0)3J@ZxcZffCr)^T<9pEU8MmjoNwQcL&{q`
zq$C(DK_Y0*_648F<*c)1)6*bCLCHTzq=+6-06jGgLP(SpP*6b2`CeM5NO97B^WDzQ
z?Ckp5_PNJ10IRn*od-KUX<(;y*Y6}r+Kmo_VYsl`y63cZ{7x?=&Q|t=QN#$&s4Z}A
zs4<37A#79Zp+s}0g|rPV4HZf-Y#UY>OtVd7>KtAeK@8WD+eYLDGj6z1Q$|jV(To@(
zrjAj$5L8hqh%Iz3uu>GWDPbl;62>W?s<>!H6U8;w$`VB<!U$uOW|rWE5R&1nQA>zI
zP{9-x#u73Y+z6sDv(He~4$6hF&(OJGL@}+E-G?b|kz*x^vhR007c!fp72GX?QVL9s
zJ;A-QIh>6=n|bq=b7#|EPsYP2EM5|6P-kPyUu@?w<htWD==Z}EwSz&_ad8$I_R=uu
z4iHD~2vOw5h-BfXh#)W8A9!(;rb+5O2@jJv9C$?{FOCiJy5Y#%?|PDDSq4G8SQU`1
zhe>iUaGE>Tx;b5#DJrWx8b#q4xjhs<3*!Dk*k}jw>3j`xI%B;)D`&ff+=a^kFq?go
zHw4FjKV6(F0mIdoALlC~@EUK_M8RM7Yhv)GyHpd0{{2x+0+z+nTtyNZH@?kQq~P<1
zw>4>4xfIo8;QRUGnX)hn<FzjfB``SJdhV3K;oAF~6$m(aaiszYztW!<N>JcDtc`|$
z*T2q}VBpniSiv@&tkk!$3-5mXuJ-J~qoY6YAMlm5?2rTi03VA81ONa4009360763o
z0J#IT+B=TqR1^kace4ScNvB6_V9<F#27)P!)FgvYmjeVrld}mD2-pfDLp){`?5L2c
znvpF)Hl^Q{Ir((--*X?nu7~0B`{m`U;rWT5-3;%apHEM}VHk$v!-J5gw}05i-tl-C
z$NT$fnhwKv?+wG3hw=8!r(ao4GWqcNoALJer2hKy)o{12x|S2?jQwWD`qivSX%w6d
zBGvV3ry2WVon@{JQdxB$p4{KI4Vea!_e}j_rrNb?P)EqBdYq{nqLnU}NHiOso$6>d
z*@dLwY*>-6v(t>-)L%kvxr3V>p$vDGxXpZX=l;I2b6a~cdtc8~y;jYX7F|`3Gj&5N
zaB@;Nnl%c^DmWWv#tCODy$znNTZ=TD%}&cj*Zw;@&Dbrofha0c@Gh4a7u;1g)e5|8
z<jt~q{{FFT{#sY-ycuYw%C)MS7a3X&Ez4L{k27`4S`(JUgl2_Uy&cX<W?R8o)i4q`
zYg``iAyZQ!hi7XU9R#n|$%XJCxQnwCo-ND8W`<!Si5$*`6sy3q4NrB$S6gmTie^)}
zqYTcva5=oeia7(nFm_faHJmMT%k@r}X6&|OO{rwTU8WZ89b7SOR=k_+%S#1}t4z7Y
z{rd>=cH0qziI-OuAIwz#N0YtXcmxskCPuR&DedrT#ZyTM&8l%2IhwWTwrn&T$|Q>y
zUkq#Pve?uyY^p71FN966QioTYVaf_;%dHt7LWgjqt+LaM-L^w8$KoJ{aoO0y8SV;K
zmF(bZx+y*f=_uJDUlnVyJ=|8Q!n;N(zP*3H|MBDD#p$DU*M-vgaU2i#_tP{^!|R)O
zF3W+&SWLICfBs~+I~||hJU#!{EN)^`c|6>o%RP(>cCrKOVq^c|r_YAxr*GF-&c56X
zPUjHD;m_N5Dvf#2xJs?!b1OhqDd!Tjs?;R}S4{$+fQHfeMIEcAl;IPox_sk<s>YHA
z?}2PdfzN_!>hqf!RBelLcmnN|_|a--J#URzHBxe2J<im-4Qm#emjs^mu=CvqmSy3R
z#IsIR_-r}aL?0(GD@<wyuU3rO1J5R@IGj~26~(g|wqaNJzpOcn;;Xd~cYL;?)nRxx
zS*q}CXE=Vyvbzeu)RYwM(6`-f0nb*5Ylk5d67e0X<%kG|v!a~%6->mNnfG8l6V9qq
z5S}$^Yr?ZE)s1H*xbQ<(?V|W<o7HLuHskWgEihYr`A~#rMH_yqQKG6`Xa53j9M$l7
iDgXc<iwFb&00000{{{d;LjnLB00RI3000000000a8~aoM

literal 3268
zcmV;#3_J55iwFb&00000{{{d;LjnL)0#%YZZ`42(fFDFjaOfzY9i;&gyq^0oQf_fb
zNeHY2iKzB)h(#Q)?A0zk4MG%@{DVY_=n)0bQ_~=XL`eY!1+>ii(yS#*{@&-A_ujl&
zU)w%+eFk9l&ZhTpCnPQGweN+!G|l?)ac3AUthVoa?VYf9m=SNw{$La_f-~w0To`JM
zVN?j$lxrx_oM|ClLn}jt5)8YBl?GG0h)m_-r4ht%ExBt%VKC!{8#QI*#2C$pAz~_x
zDukelN<mzq3xSoQm`w>Y5t1-Y`BcSaE1D>-u~v>KI}t`0qcn2_FNKf{+eRHBN<jrv
zR2WCdLU1F9!pzN~>J^j=;pWhVU_>#kmD`6YPmyCKiE__(yAaaO(FyLAK`8~M#+~4P
z)f~1Xw=>_p?cLoB*VFMZj>?-vTGZRv3KzRY4Eg>z>kI}_hPs_R?)lh8hKE_y>F0<e
ze}pLV6GUwIDIzF{5Aq<1vn<VmC(&`5M0rqd6eNj3K|dM=2mL@YyG!0l7V84+_b^S5
za&LFX`QDu_%#@YY9*yE?jQlS0Q)FX<qo~#GB*@PZZejZf6g_Lg3zq?4*1nS)0jGaI
zU7V``!_}7`=W8PH8gDd2AzTg{V(_NF)DVZkgHb~Qmc`OsO%hr+zs=U9;PZ#K4QW`p
z6gOnx`}xzEsxS)UwJ!@5FgV$I?p46y+WT8I2snFjr3MMVvY!_!P~bgkjD~+VzRp)*
z;MHnW!#13)G^eo#?|%HQ*X+aNlRxkuD)$4nUIYLDABzYC000000RIL6LPG)oQVOM7
zTZmm(8QwF~n0BJIO>9zh&Sa`KX}pAWzf`K-U4`A2B<8Z!v=vXO1&v?{^raR>`{t7t
zv{cZCJc)>xAZQU0s|I~&9@-Z*4^p9s6$Ohvh}z`%t-beJd!KZsYxX(^_GA)rKK|?X
z-}Xm}_Ve;5&R(}ZTQy6?4b9(Y-N*la>f}jAq?T45rc?>Rl%3YKtE###U2$%uC{CAd
z`RLsbD92SzMRDw?TRwp=F5$W1rE}LzW*6U`0LD+Lq@0#5ZtaSvmKxk8;4<#02VMGb
z2YTY4zXE7|YXa0#A+46vstTawng+D+!fFp$9LHCd@x1iKPZsO5CzewN;G9dT#T2S{
z26#Q@i4ODQe|dU&GCQ}n)ERjTOsFme;n2PEkgg7pUUzAoj=Fo$zdp)$OlErzCP0nW
zgb5NHRn?ow86cmz{aHZXdME*@2&Y6t&eoi+D$XuPNN(pKi+dh?(`5FO_oh6=n5A4#
zT~#+yR>iA5r0Atw!E@=41+01G?I~+2N+^?4ueWzlRUBW5kdjBId&n1lw6U^2dwx9u
zNf{T!SUauC7IQ0O;vb9Imv2LC{_UL!OlGX0%1kR)HnCY3e;R168pu-}<gfQ00p#c2
zm4JkhmT5Ds>!A6@Qp`<CN?A2ov>YV8!jHj6Po{iC8DNP;cs7!jWpQ>XO3HXteg~er
zK6?MsWcGu7so+EoQOc&JYuxV2;;Dg;^c>_9KmN?hWcIK92`RNOS`oy16_V)qAUNkC
zAHDE(Kz?f@C8bctDvn5UTe|3Y2l>r^d=HWI+D6KjL}_U_1%0#u|3=0<GyM(3eE60G
zrs9%O%A<ek#=9XnMFV*fU)hQ0cmFF^)@Dc6Q)x*A@f_D8!abxXc>$KMNDL&8`t)}%
zAwGYYqGKb7GE7XXHa?;ItG(A%pP3%=`n8|$SfAZ-C>fhpGRsWFthUI^Yw_$_dl&@r
z;yY5A$psfyAUMmeYCJRuPB900?__(BOoNLxn6vIO^ZX!^_#EWlw+G3E5QO@8Z>6j@
zNud)R<erUfK`JV&x21D|-v@O_)r!PAbRC}OuWSJHVv3FpJS4RW)~wtX4)Tzv9^ru8
zo05}KtC-KrmW>gfX9udwImicYdIULiX@3fm5YDL%)`Wvz7^J0*)?CH&*$;fMSeuo5
zQiQ}LS_SZXJ>MsIV{%h6(^&Il2YTZ-9s_aiyE)k=DaEZpAeHkSP17bxC;C2V7oJ}{
zdAXR(KD0NXy=ZzbFnn7w>v9VpdB_*v|Gu4**?$hF1hqzJ%b>pNNa@Z29jiIWGY|f9
zbuv44N3!)2M#4v~YNgyP)xA02jOS~w?1I}LUr!bdV$dzX?ZHW1N7I;-I@-2g$2-*G
zas^WQ^uc7iASL3-w+o@RKR1Y`#_>hnVV>FZGxT2aP|8JwX{~$_uID?DhLV|Myd<&j
zZg5X~T$LhFp{(VCg>I>y>pgWPZ8X4Jd?WIO&o8adZofHg0B%~oi%CDoP#V4OF<1WZ
z43hpkYpH-T)F|cq?iRA#!uuZb_tSr)u{m%!MWNPGUpT`oQx)g>nS(M%c}%A3@%;Cv
zQ>5r`_ay{nMo5ZYF;GdpN&k7GgZ$Wm3)fC&_sfK!mQam!iiC%X)C)>#)+ha2JwNSN
zNcvm$BtTI&74Z=j8>Z(5HAzG{J>*wifnd7tKnl_jZHx+>2e+Rc3~{U?b8_Bv+xNjF
zmr1HBDI_TxyKd@0+MgfT(?-X8$P2Hnz@DdBs=Yz?>1$Hg-(1C_suaDT9!I^r3xYnL
zqCCO4vXW1$HkF|Lmj;6Nls5v>a!Vu;aXczFiKUYr<g*W1=zf@|%xM`X=;ph*18(2w
zDUaH+9p4+>cjLM5N`X{5@}89Em1Ir#-FhAs<otl~`bpbE?tJXVo$Ir&98R}(j9X?y
zYX{78g8*gGo_l<NqJ~ap_wG*3hD>wI(JGf=D1Lp+d&sx>?%~g$Ld?H%FxAvb0eg~Q
z&$8NL$KWBqeenwF$K5GoV#gq<VJxQRW<ArLgYT~u_v~&*x~(xKYs+ycAE%GH)sYyW
zHEHRflgq$RMiS{UT)rI2DvQp_m0`>UMo|v_|708XAWBfoBCZ|^G8)7P?uU;yJ!_XG
z3AMs<5@z8{Ntnc1&J1S297zx&R4Ev?><v3!5=)d{W_3}Lrw+H2AoV*z>O6Ij>QwaO
zdzl3yjli@K)(Ul=IvCLf14q=u%oz=0(Wgs*xNhpn0nzv{8C{$wiQx)EatO9O?=T<t
zP41NWxx*-@iqas7%kmBr8ZMTDm3EFKibzgam?T`431YCl4CTpXB_Pw-&<izh*wG-^
z07=dk%i?A?gQC+Bb)I)P5!BC8&V{8>hG99;ihELOw{CToVL>KJ847j6t~v*V8Bnzi
z`@cM*=CCYsa9Y=OPX(hvIAV?3FjQ1|l1NL4X5nm;M@n323O9%CdS0+W8~ka*iY+fr
z1fxPXOB<ITvBD75Y<}YQGz-L9v^fa-7Q|(NSc#@j2X*p_m+#jY=&_1xmn;yC)h|vH
zuFP+80Ahp`@v@U;6Ui`8k;rT=i|s*;ocpC_xj^xf0yAhcFS-Rla5=<TF3@8hpD2M<
zh}Q7{nH90hQfP$jaa}JU3Br|N_wE+)jwW2;6C}x_M~%&pR63LlSLcCf%XL38<w?S@
ztx)~NIm<hmQ)+|>%Za>T)7)^2u<vh{IUtz3VA@at^Y}!2hHVCfdI6hg!8L5st-bT=
zjyBlNQ5m)rd10@u-|y1-GCCH-5NgnIg?T5hcp1V7R0_G1*A*CwS}bXFvCEQ#V~MQe
z#=FXjSi=n_I1&$Uc{UOFP5SYz&ZkElli^_+m@U&L2v;lOsW%e@JfIP5)zHObL9Dk(
zW0M8C0*i4}STjFmjCHYPSUyq_hNe6hTZS44?{xdiJU+p?!59qJ<&daHtAjPF^gY{f
z8XYZ(Cd^NxRk=W0Zt;^?8eS)<sw{Pk=4h!0?{#H?7~y-ig>nZSj#B-Rv#rIecMZO<
zzMYL9MIX!JOxuO*XcUHnGg0w<qB0W&_fY@xRJ>nhX@s*smpZ;RROg{+2!o!hBigQU
z8;xR!RRXO}H7sFcQ7FL%#K&CgX;)>UApEr)T(GlH;2`3wSh)}i_pfG%tJb^YQILzA
z#aSWGN7#a3OlDCdm#2}$FaEMs4tZfbipKOKLd#Q?c_<h;uy|Y4E{0Km0xR7@v5SkB
zL{IAGYRFg=Mw9~W#v+o5f&qv^BUcXE{m~jRfq7fS(ZAb_jz<AYHC(p{?=tLWh%Y2n
zxd1aTsVQz@VIxvzYNWV7N5-{trghP<3&lfe%trAqfy8=L=V`>z5H|0I&m9-WqZo~o
ze{=ik*Q}#a7^Y?=uDWH7M&W33jSlgWX<dKL6-EMO5oUmcu+6h!4N&K007tN>=6LEB
zsMAmgwL}eGE<b(Lt^OY}&EBU>FaQ7_iwFb&00000{{{d;LjnLB00RI3000000001Q
Caw19q

diff --git a/src/umi_tools/umi_tools_dedup/test_data/sample.bam.bai b/src/umi_tools/umi_tools_dedup/test_data/sample.bam.bai
index bd87cde81a00c7699faed37ab04f88fc5f77bcfa..e9e2eee1d579c04b0b58bff33862d56ff931ff5d 100644
GIT binary patch
literal 2656
zcmZ>A^kfucU|?VcVnbmD21X#wz!1d*B8?;=wAUmC2Cx8BoPj}v4J2O+B9K5ZRDTy#
z-$N*E4Hbu(2b0f+io?{w%vA<E0%AAJEu-{kctDccXgG|911JPW)4^yu7)=L+@&T;Q
Wr4|jVyN4<bYikTu8j|OS4haD7sht1-

literal 889032
zcmeI*L8zQp6$kM9&CGk)G-;9sUeffDrpA|$1|CwWs1R(6SULph!h+cf1)(5FL7LKq
zA}-uW9bAb8p&M}#cY>=z7VRP}Asf++tD=JF#wrL6cyG@=&gX+HViTCj`+ble=FYv}
zOV9kz?={V1=GkYSAGS&<lltqx7fKmwE#<}buw3qSzKG}duE*ZCGp=hpt;ShZA0id{
zpbuQ&AqP0X0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K0
z2ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8
zaDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h
z00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70
z103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<6rogA3f-|T!*
z%B0rIC0@H>h}zc1*L$BSWvC0e?`q%w^_^JH$9ewXvE|-8qegG_K_A*|$$=cmfjQs+
z2ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8
zaDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IB@?vaJ59vRefmlE(daOCpp-x
zn}4xg%24ZzL-g)fz3!Zk^;`Y%3zK+$>s+kozI^^S7h*m4z18=>Hi`A2-zUG$d{Ih$
zs^og^-)R2=qwCv`EcfJIHC_vS(1$iEav%qCU=BFI0S<7019!)P{IiN0{nQ72Xy0Gt
zKn~=<9B_aG9N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02RN`!4&-}fs*ziL(1$j^
zav%qCU=BFI0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K0
z2ROh14(wkCZVi!tRUeuIav%qCU=BFI0S<70103K02ROiiyXn9Sljx_a4=q=6AO~__
z4miL84sd`29N+*4IKTl8aDW3GxEBsg>rLm2QYN*&)&HO5d>XxOYvb$ssZxfz@UK1l
zE}e|^JfF}1;NbF}`J5WP)dziOvn2;|AP44v103K02ROh14sd`29N+*4IKTl8aDW3G
z-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$
zU{xH*_X<$MC4JC`=Cd5gfgG3v4sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<700~}a&2WB;-SI)%ehP>H7Z{)*Xzkeat^YL@DnCZWMI@a^^N%CBNeo3z9{`T1>
z<Lir$)XzHUUqg+a>VrPC8Ic1ykOOnT0S<70103K02ROh14sd`29N+*4IKTl8aDW3G
z-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$
zfCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCKy8fj6h&fT|D8LphLx
zdzpg|yEkg&Ods^2&6^y^fgG3v4sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0fpv3WvxfGY
zkC!sk`qPb=;Xn8KkH=#DXR}!UTd(s)DU-VYx&Ao!Z?pjzT|Yk^=kwgt{k|Xd^?aOP
z_v-Q3pZCqLd-Cw|Yvprlye9gf4{hY+Kn~=<9B_aG9N+*4IKTl8aDW3G-~b0WzyS_$
zfCC)h00%h00S<70103K02ROh14sd`299Vw`@@Hq&m}h;^hvu3b$blS~0}gP2103K0
z2ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW3G-~b0WzyS_$U=<z6&t_92pZcH=ZBFGt4(=oe`B|1~<VGL%S0C~-
zT-A8p^+6xn_mA7<pgq^FS^brtr<|Xo{K01QZu;?#<v%W+iqA*ReHZ(_i|1qgjs7^#
z|9U6Zb6-CH&#hAO9%}SjAM~M3njFZ19GC+RaDW3G-~b0WzyS_$fCC)h00%h00S<70
z103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh1
z4sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROiiyX`=JW`G*L
z=z~5qXXQW+<iH$ofCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S<70103K02ROh14sd`299TUE-q?&>s`}98Qx4=n4$J`uIKTl8
zaDW3G-~b0WzyS_$fCC)h00%h00S<7019#%Uq;9_XXemRjAN0R_eDYJ#!}+3=_Wa)U
zS^EN`>qieP_vBtRdZ-Wj&_+cL<UkJ00S7q10S<70103K02ROh14sd`29N+*4IKTl8
zaDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S>IH1MR)V{_t45AKCBw`;c9DAZ9k-k1XGxEZ3(k(4*_UlW{)x
z{jlF>tG~}#KF<69<WMQOSB+llgFdv8k^?!A19QLu4sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h
z00%h00S<701FPadzGsLUF6qN3>qEZxj2eB^2YqPY@8v)a<iH$ofCC)h00%h00S<70
z103K02ROh14sd`29N+*4IKTl8aDW3G-~b0Wu+9#&&oX=FSbTn2e!kiF_I&<XuK%vD
zPujom=sJEP&gZ$@cVnZJJgdf>=z~7A(UAi=kOOnDE)L}Pb~W762YqOM%YhuofjQs+
z2ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<7019#tnS#9dScjM>M
z&mV}4z2575QA+z6_W%BzdS7op3qQWTb}`Q9xf^}om$za)&*%L&FD&nw&#BRSeb9$C
zTXG-=a$pWPzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4?zIE?clc^}qYwJf
z+?4}4kOOnT0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K0
z2ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8
zaDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)({~UPzK%7(cp*boCav%rhfCC)h00%h00S<70103K02ROh14sd`29N+*4
zIKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G
z-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$
zfCC)h00%h00S<70103MM`Z};tU%XdOl`_<Nqkm@Q&V!NZZ=H+ve0-&^&ljakM*Ghm
zjQz6~gwgfX`8c2FZuaw^@9X(E&;NaE`FqIUiyFPv2YqOxB?odK2j+kS9N+*4IKTl8
zaDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h
z00%h00S<701MA^He%6B;ua`dPL;L<E2XY_>=70kn-~b0WzyS_$fCC)h00%h00S<70
z103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh1
z4sd`29N+*4IKTl8-2V<-c`$OW>O-4%IgkT6Fb5pq00%h00S<701E17^oBhw3vl^yf
zZkICD`r5&mt@nHV(ur6<+aKri^F=9>x<A*?9Ex+(7R=Fgb~n!F{c_(QA6=f!d#KSv
zeb9$CLUJGna$pWPzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G
z-~b0WzyS_$fCC)h00(a8K>lpM8cyhgJ~S`oKn~=<9B_aG9N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S<70103MM?HtJW#8krxeb9&Ir5wnC9GC+RaDW3G-~b0WzyS_$
zfCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K0
z2ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW5%j|0>CVqSi#l%dw|9*S&V?{&T?Wm4DA>{)MX<LlkyaX$CueU6@4
zp3Udf=&?TNLz^KvkOMg|2OQu42ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h
z00%h00S<70103K02RN`Q4&=`OtKpJ9=tJ{a4&*=%%mD{DzyS_$fCC)h00%h00S<70
z103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh1
z4sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4
zIKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G
z-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$
zfCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0Wz=8E~V57cR-*~c=q1JaEitKJ5
zj{5kqSigQK9-r#Z&(FqsK7J+F7o|+<eR6-^XVwBSx^8_o&gZ%Ox<^hg&*nYU=)FGZ
zLmMGEkOMhb{W-|*jq3mYw}XD@NApV#<UkJ00S7q10S<70103K02ROh14sd`29N+*4
zIKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G
z-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$
zfCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N@ryci@$W!U0tunul^A2XbHzIKTl8aDW3G-~b0WzyS_$
zfCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)Z4-UN3-<MU5yz7HL
zw7Hi9IgkT$zyS_$VEr7J))(#7-BN~Hw+}~#uk|`#lrkBupEwfFw=?5w_j7SR&*i>9
zonGEE@2y4;^+6xn49S5U$bmWF00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G
z-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$
zfCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K0
z2ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8
zaDW3GSQiJb9gcZb^`W^X2XY_>=70kn-~b0WzyS_$fCC)h00%h00S<70103K02ROh1
z4sd`29N+*4IKTl8+&c$aZ`(DkzvhclCbgb85}A9u*S0pkE<I7oP#1FF%YEOEcVj&t
z=lQR0FZbpdHF~TM`p{-e4&*=%%mD{DzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8
zaDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h
z00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70
z103K02ROh14sd`299SI(UOp0eRP~|Fr5wnC9GC+RaDW3G-~b0WzyS_$fCC)h00%h0
z0S<70103K02ROh14sd`2>+is%zKG91T*^@E+x=%l9zGh`o-ayi&+lDtd&k%K?6f}d
zmveEyyw6K}?w9Mo?d$DpjIXae5$E$<?)&I?DS1|n*GM1qq0NpQ$blS~0}gP2103K0
z2ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`2
z9N+*4IKTl8aDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8
zaDW3G-~b0WzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0W
zzyS_$fCC)h00%h00S<70103K02ROh14sd`29N+*4IKTl8aDW3G-~b0WzyS_$fCC)h
z00%h00S<700~}aY2VOcF^Qr1Xb4m{6Kn~0S2ROh14sd`29N+*4IKTl8teXQZv+ep$
Gl=2^72Ijl~

diff --git a/src/umi_tools/umi_tools_dedup/test_data/script.sh b/src/umi_tools/umi_tools_dedup/test_data/script.sh
index 534c4af2..2253a0d1 100755
--- a/src/umi_tools/umi_tools_dedup/test_data/script.sh
+++ b/src/umi_tools/umi_tools_dedup/test_data/script.sh
@@ -2,6 +2,7 @@
 
 # Download test data
 wget https://github.com/CGATOxford/UMI-tools/releases/download/v0.2.3/example.bam
-samtools view -b -o sample.bam -s 0.00005 example.bam
-samtools index sample.bam > sample.bam.bai
+# extract 150 reads with a maximum of two reads having the same start position
+samtools view -h example.bam | head -n 150 | samtools view -bS - > sample.bam
+samtools index sample.bam
 rm example.bam
\ No newline at end of file