query regarding bedgraph output

[pfa...@gmail.com]

Hi all,

I am a new user of quest and have a few queries regarding the bedgraph output. I have uploaded two sam files gfp and oct4 (mm9 genome) and using recommended settings 2. the following info is:( earlier the default chip fragment size reported by SICER was 150 approx), Quest has reported a peak shift default:60 from the following info which is slightly less than 75 (150/2 ?) which is not offhand I guess

used_regions: 200
peak_shift_estimate: 60
estimation_method: mode
 
further  info:
ChIP peaks: 37852
ChIP peaks accepted: 21953
ChIP peaks rejected: 15899

ChIP regions: 37036
ChIP regions accepted: 21953
ChIP regions rejected: 15083

i.e. without fdr due to not enough reads in Rx NoIP..

In most of the tag shifts(+ and -,bedgraph files) i am getting values mainly just 1 and  2,  The loaded tag files are just looking as separate blocks in ucsc and igb?
Except a decent profile in chr1(just a site) and chr7 particularly IQGAP1 gene.
On the other hand the peaks and the wiggle files look outstanding and smooth (maybe even better resolution than MACS or FindPeaks) and getting very good FE as well from chip out accepted.

Am i having this problem due to pcr dups that i need to remove using samtools ? or I am missing a library which is only affecting bedgraph fusing step? My quest version 2.4 compiled in ubuntu 12.04(Vmware,32 bits)
what  could be the reason.
Sorry for my long mail. appreciate a reply.

suman

[Anton]

Hi Suman,

I am not sure what is the issue here. Are you saying that the peak shift estimates within many individual peaks equals to 1 or 2? QuEST is using a single peak shift for the calculation of the enrichment profile (60 in your case), and then it also estimates the peak shift within each peak/region, but that is not used for anything except for filtering out peaks that look like they are artifacts. It does appear that you have a lot of failed regions. What threshold are you using for calling peaks? Are you using SAM files for your input?

Anton

[Anton]

Thanks, these are warnings, not a big deal.

In regard to your earlier question - you do not need to remove duplicates using samtools, as this may erase signals within the peaks. QuEST already implements a type of deduplication by default. Regarding the bedgraph  display - it is normal to have values equal to 1 or 2 within the weak peaks.

Anton

On Sunday, January 6, 2013 4:46:37 AM UTC-8, pfa...@gmail.com wrote:

In continuation to my problem regarding the bedgraph output from sam files. I got some warnings during quest 2.4  compilation
in Ubuntu 12.04, I want to check with the forum that whether they can be sorted

warnings:

params.cpp: In member function ‘void params::optional(const char*, const char*, const char*, unsigned int)’:
params.cpp:117:16: warning: variable ‘par_index’ set but not used [-Wunused-but-set-variable]
g++  -Wall -ansi -pedantic -O3 -funroll-loops -pthread -c string_utils.cpp
string_utils.cpp:248:3: warning: integer constant is too large for ‘long’ type [-Wlong-long]
string_utils.cpp:249:3: warning: integer constant is too large for ‘long’ type [-Wlong-long]
string_utils.cpp: In function ‘long int long_k_pow(int)’:
string_utils.cpp:248:19: warning: overflow in implicit constant conversion [-Woverflow]
string_utils.cpp:249:19: warning: overflow in implicit constant conversion [-Woverflow]
quick_window_scan.cpp: In function ‘int main(int, char**)’:
quick_window_scan.cpp:263:10: warning: variable ‘cur_max_ratio’ set but not used [-Wunused-but-set-variable]
collapse_reads.cpp: In function ‘double Poisson_p_value(int, double, double)’:
collapse_reads.cpp:40:10: warning: variable ‘prev_sum’ set but not used [-Wunused-but-set-variable]

but quest is running without any warning !!!
however, may create any problems with the bedgraph output ?