Erange

[Kasycas]

Hi,

I have a traceback error as well but it's not related to Cistematic.

python geneMrnaCountsWeighted.py human bowtie_1205_2_hg18.rds bowtie_1205_2_hg18.expanded.rpkm bowtie_1205_2_hg18.multi.count -accept bowtie_1205_2_hg18.accepted.rpkm -multi -cache 1


returning 37279 regions
human cached
caching ....
dataset bowtie_1205_2_hg18.rds
metadata:
bowtie_mapped True
dataType RNA
paired True
rdsVersion 1.1
readsize 40

19380512 unique reads, 35893 spliced reads and 1531861 multireads
default cache size is 100000 pages
found index

1 read 100000 read 200000
10 read 300000
10_random
11 read 400000 read 500000
12 read 600000 read 700000
13 read 800000
13_random
14 read 900000
15
15_random read 1000000
16
16_random
17 read 1100000
17_random
18 read 1200000
19
1_random
2 read 1300000 read 1400000 read 1500000
20
21
21_random read 1600000
22
22_h2_hap1
Traceback (most recent call last):
File "geneMrnaCountsWeighted.py", line 119, in <module>
for (tagStart, tagReadID) in hitDict[fullchrom]:
KeyError: u'chr22_h2_hap1'



I get it for another rds file I was running as well but it failed on;

"KeyError: u'chr16_random'"



Thanks for any help. I'm really not sure what this error relates to and how to fix it!

Kasycas

[Howie Goodell]

Hello --

I am trying to process an Illumina ChIP-seq experiment with a background: two 4-5GB files with about 30M mappings apiece (ELAND_MULTI files). Using a sample of 1M records, I got credible-looking results. For example, the output for chromosome 4 in the sample run output looks like:
chromosome chr4
calculating background...
5 36.0783776167
Poisson n=114533, p=0.280103
#regionID chrom start stop RPM fold multi% plus% leftPlus% peakPos peakHeight pValue
s_2-s_19 chr4 80274411 80274599 6.1 4.8 43.9 33.0 67.0 80274448 3.0 5.1e-07
...

However, with the full files (ca. 30M records apiece), findall.py runs for many hours; then fails with no error message. All the chromosome backgrounds are 0, and there are no peaks reported; also the chip_regions.txt file is not created.

Procedure:
Following Alli's helpful responses to earlier posts, I converted two ELAND_MULTI files to RDS using this syntax:
python2.6 ~/bin/ERANGE3.1/commoncode/mkrds.py $1 ~/test/$1_eland_multi.txt ~/test/$1.rds -index -cache 2000000
(where $1 is the unique part of each file name).

Then I ran findall.py as follows:
python2.6 ~/bin/ERANGE3.1/commoncode/findall.py $1-$2 ~/test/$1.rds ~/test/$1_chip_regions.txt -control ~/test/$2.rds -listPeak -revbackground

I don't believe there were any different parameters, etc.; the difference was just creating the two RDS files with 1M source lines each, or about 30M each. Is anybody aware of a limit in the program or python (2.6.4) that might cause such a silent failure based solely on the data size? It was a 64-bit computer with 32GB of total RAM; other processes took some of that, but I don't see any obvious external limitations.

Any suggestions will be much appreciated.

Thanks!
Howie Goodell

[Kasycas]

Dear all,

I'm still having a great deal of trouble with ERANGE. Whether I use paired end data or single read, I get the following error on the very last script of the runStandardAnalysis.sh pipeline;


/usr/local/commoncode/geneMrnaCountsWeighted.py: version 3.7
merged 0 times
returning 0 regions
dataset posStrand_read1_paired_hg19_spliceCore_ALL1205_10m_1mm_hits.rds
metadata:
bowtie_mapped True
dataType RNA
paired True
rdsVersion 1.1
readsize 40

11744838 unique reads, 0 spliced reads and 0 multireads
default cache size is 100000 pages
found index

1
Traceback (most recent call last):
File "/usr/local/commoncode/geneMrnaCountsWeighted.py", line 128, in <module>
for (tagStart, tagReadID) in hitDict[fullchrom]:
KeyError: u'chr1'
/usr/local/commoncode/normalizeFinalExonic.py: version 3.5
reporting fractional contribution of multireads
dataset posStrand_read1_paired_hg19_spliceCore_ALL1205_10m_1mm_hits.rds
metadata:
bowtie_mapped True
dataType RNA
paired True
rdsVersion 1.1
readsize 40
default cache size is 100000 pages
found index
returned 29469 genes



Please, has anyone come across this? I'm going mad as I can't see it would be the data I'm using, nor the way I'm running it as this is fairly straightforward...

Thanks,

Kas

[mceachin]

I realize that this has been posted before but I still get a cistematic error

>python2.5 $ERANGEPATH/getallgenes.py hsapiens chip_regions.7_vs_6>.nodirection.txt outfile

>'import site' failed; use -v for traceback

>psyco not running

>Traceback (most recent call last):
> File "/usr/local/bioinf/ERANGE3.1/commoncode/getallgenes.py", line 6, in
> <module> from cistematic.core import genesIntersecting,
> featuresIntersecting, cacheGeneDB, uncacheGeneDB
> ImportError: No module named cistematic.core

I'm on a 64 bit linux machine, so I don't think I need psyco. I've exported the paths for both python2.5 and cistematic.

Any other ideas?

Thanx

[EGrassi]

What directory have you exported? I had the same error when I mistakenly the path including the cistematic directory (cistematic has a subpackage called stat which clashes with the basic python one in this case and causes the import site failure).

[mceachin]

Thanks, I'll give that a try.