CONTACT
name.su [email protected]
Resea ch G oup Sys ems
Biology / Bioin o ma ics
Re e ences:
[1] Su name, N., Su name, N. N., and Su name, N. N. (2008). Na u e Re iews Immunology
Pic u e o you CONTACT
Ruman.Ge s @leibniz-hki.de
Resea ch G oup
Applied Sys ems Biology
Re e ences:
[1] Anika Klingbe g e al., “Fully Au oma ed E alua ion o To al Glome ula Numbe and Capilla y Tu
Size in Neph i ic Kidneys Using Ligh shee Mic oscopy,” Jou nal o he Ame ican Socie y o
Neph ology: JASN 28, no. 2 (Feb ua y 2017): 452–59, h ps://doi.o g/10.1681/ASN.2016020232.
MISA++: a s anda dized in e ace o au oma ed
high-pe o mance big olume image analysis
Ruman Ge s 1,2, Anna Medyukhina1, and Ma c Thilo Figge1,3
Ligh -shee luo escence mic oscopy (LSFM) allows quan i a i e h ee-
dimensional analysis o whole o gans. This includes he e alua ion o
s uc u al changes such as a educed numbe o glome uli in kidneys [1].
4. S anda dized da a managemen
1. Implemen ing a big olume image analysis
1 Applied Sys ems Biology, Leibniz Ins i u e o Na u al P oduc Resea ch and In ec ion Biology – Hans-Knöll-Ins i u e, Jena, Ge many
2 Facul y o Biological Sciences, F ied ich-Schille -Uni e si y Jena, Ge many
3 Ins i u e o Mic obiology, Facul y o Biological Sciences, F ied ich-Schille -Uni e si y Jena, Ge many
Whole mu ine
kidneys
23 kidneys
110 GB
x
z
yRegions o in e es
e.g. Glome uli
Quan i ica ion esul s
e.g. Volume
41 min
3 days
4 days
Big olume image da a analysis is ha dwa e-in ensi e and equi es high-
pe o mance implemen a ions in e icien languages such as C++.
2. MISA++ - A pla o m o cus om C++ analysis ools
You
so wa e
1. C++ amewo k 2. JSON in e ace
Pa alleliza ion
Modula iza ion
CLI
Memo y managemen
Pa ame e s
Da a o ganiza ion
Da a ypes
Quan i ica ion esul s
Easy de elopmen
Easy in eg a ion
Easily ex endable
Ready- o-use componen s o common asks
S anda dized JSON in e ace o da a & pa ame e s
In eg a ion o cus om da a ypes & mo e
Comes wi h:
3. Pa alleliza ion and modula iza ion
6. S anda dized pa ame e s & documen a ion
So wa e lib a y
Task Task Task
Task
You app
Task Task
Task Task
Roo
Task
Wo kload is o ganized in a di ec ed acyclic
g aph (DAG) whe e nodes ep esen asks and
edges ep esen dependencies be ween wo
asks.
MISA++ au oma ically uns asks wi h sa is ied
dependencies in pa allel un il he wo k is done.
Easy modula iza ion A se o asks can be
sepa a ed in o a so wa e lib a y o be
de eloped independen ly. The DAG o any
MISA++ applica ion can be e-used by ano he
applica ion ia a s anda dized modula iza ion
in e ace.
Easy g aph cons uc ion MISA++ comes wi h
op ional unc ions o simpli y DAG c ea ion and
enhance code eadabili y.
Task in DAG
Th ead-sa e
Read/W i e
Da a cache
C++ ep esen a ion
Me ada a
Folde
I/O
MISA++ w aps cus om/ hi d-pa y da a ypes
wi hin s uc u es e med caches.
Au oma ed I/O A cache is associa ed o a
loca ion on he ha d d i e ha is used o s o e
cu en ly unused da a. On accessing a cache, i
au oma ically loads he da a om he ha d d i e
loca ion.
Th ead-sa e access Tasks can access cached
da a ia h ead-sa e unc ions.
A aching me ada a Any cache p o ides
unc ions o a ach use -de ined me ada a, such
as quan i ica ion esul s.
5. S anda dized da a loca ions
VFS oo
Da a cache
Da a cache
/da a/imgs
/ esul s
Caches a e associa ed
o a loca ion in a i ual
ile sys em (VFS)
de ined in C++ code.
VFS loca ions a e
linked o ha d d i e
loca ions by he use .
Flexible da a loca ions Use s can ei he ollow he VFS s uc u e o
edi ec cache loca ions based on he cu en machine's ile sys em.
Inpu
JSON schema
Pa ame e ile
s uc u e
You so wa e
Pa ame e ile
VFS links
Samples
Algo i hm se ings
Gene a es
MISA++ applica ions ead all applica ion
se ings om a s anda dized pa ame e
ile ha links ilesys em loca ions o
caches in he VFS, de ines samples and
hei pa ame e s, and allows o change
algo i hm se ings.
Any MISA++ applica ion can au oma ically
gene a e a ile in JSON schema o ma
ha desc ibes all a ailable pa ame e ile
p ope ies.
Easy pa ame e de ini ion Pa ame e s
and op ional documen a ion a e c ea ed in
C++ code and au oma ically assigned o a
unique loca ion.
Au oma ed documen a ion The JSON schema desc ibes all
pa ame e s, as well as he VFS s uc u e, including inpu and ou pu
da a. I can be ead by humans and also pa sed by o he so wa e.
6. S anda dized g aphical use in e ace
You so wa e
MISA++ in e ace
ImageJ/Fiji GUI
Via ImageJ plugin
We de eloped a g aphical use in e ace o
any MISA++ applica ion as ImageJ plugin.
I ex ac s all necessa y in o ma ion om he
au oma ically gene a ed JSON schema o
gene a e use in e aces o se ing up
analyses, unning hem, analyzing esul s, and
c ea ing pipelines.
Use -de ined link
Ge s , R., Medyukhina, A., & Figge, M. T. (2020). MISA++: A s anda dized
in e ace o au oma ed bioimage analysis. So wa eX, 11, 100405.
