eprintid: 14922
rev_number: 22
eprint_status: archive
userid: 548
dir: disk0/00/01/49/22
datestamp: 2013-05-23 09:01:27
lastmod: 2013-06-04 07:21:47
status_changed: 2013-05-23 09:01:27
type: doctoralThesis
metadata_visibility: show
creators_name: Batra, Richa
title: Computational methods to analyze image-based siRNA knockdown screens
subjects: ddc-500
divisions: i-140001
adv_faculty: af-14
cterms_swd: neuroblastoma
cterms_swd: siRNA
cterms_swd: phenotype
cterms_swd: classification
cterms_swd: clustering
cterms_swd: drug targets
abstract: Neuroblastoma is the most common extra-cranial solid tumor of early childhood. Standard therapies are not effective in case of poor prognosis and chemotherapy resistance. To improve drug therapy, it is imperative to discover new targets that play a substantial role in tumorigenesis of neuroblastoma. The mitotic machinery is an attractive target for therapeutic interventions and inhibitors can be developed to target mitotic entry, spindle apparatus, spindle activation checkpoint, and mitotic exit. Thus, we performed a study to find genes that cause mitosis linked cell death upon inhibition in neuroblastoma cells. 

We investigated gene expression studies of neuroblastoma tumors and selected 240 genes relevant for tumorigenesis and cell cycle. With these genes we performed image-based time-lapse screening of gene knockdowns in neuroblastoma cells. We developed a classifier to classify images into cellular phenotypes, using SVM, performing manual evaluation and automatic corrections. This classifier yielded better predictions of cellular phenotypes than the standard classification protocol. We further developed an elaborated analysis pipeline based on the phenotype kinetics from the gene knockdown screening to identify genes with vital role in mitosis to identify therapeutic targets for neuroblastoma. We developed two methods (1) to generate clusters of genes with similar phenotype profiles and (2) to track the sequence of phenotype events, particularly mitosis-linked-celldeath.

We identified six genes (DLGAP5, DSCC1, SMO, SNRPD1, SSBP1, and UBE2C) that cause mitosis-linked-cell-death upon knockdown in both of the neuroblastoma cell lines tested (SH-EP and SK-N-BE(2)-C). Gene expression analysis of neuroblastoma patients show that these genes are up-regulated in aggressive tumors and they show good prediction performance for overall survival. Four of these hits (DLGAP5, DSCC1, SSBP1, UBE2C) are directly involved in cell cycle and one (SMO) indirectly which is involved in cell cycle regulation. Functional association and gene-expression analysis of these hits indicated that monitoring cell cycle dynamics enabled finding promising drug targets for neuroblastoma cells.

In summary, we present a bioinformatics pipeline to determine cancer specific therapeutic targets by first performing a focused gene expression analysis to select genes followed by a gene knockdown screening assay of live cells.

date: 2013
id_scheme: DOI
id_number: 10.11588/heidok.00014922
ppn_swb: 1652383522
own_urn: urn:nbn:de:bsz:16-heidok-149227
date_accepted: 2013-04-29
advisor: HASH(0x55fc36d285f8)
language: eng
bibsort: BATRARICHACOMPUTATIO2013
full_text_status: public
citation:   Batra, Richa  (2013) Computational methods to analyze image-based siRNA knockdown screens.  [Dissertation]     
document_url: https://archiv.ub.uni-heidelberg.de/volltextserver/14922/1/Thesis_Richa.pdf