[bw.]

Calling CNA and BAF for exome seq data without a control

Hi, I'm trying to use FREEC to call CNA and BAF in some exome seq samples, but haven't been able to make it work. The tool ran fine on the chr19 test data.

The problems I'm seeing are
- the tool runs but doesn't generate any of the output files except mpileup.txt_sample.cpn
It exits with the error message:
..There is no control sample!!!
You have to use a matching control sample to get adequite results since GC-bias is not the only bias in targeted sequencing

- even the .cpn file doesn't seem right because it tries to call copy number outside the target regions I provided via the TruSeq_exome_targeted_regions.hg19.bed
and not surprisingly calls them all as 0 copy number.


I'm running the tool on a pileup generated using samtools v.0.1.12. I'm using the following config file:


[general]
chrLenFile=hg19.len

window = 3000
step = 1000
ploidy = 2

#breakPointThreshold = -.001

#GCcontentProfile = GC_profile.cnp

intercept=1
minMappabilityPerWindow = 0.7

outputDir = .

sex=XY
breakPointType=4

#degree=3
#coefficientOfVariation = 0.05
#gemMappabilityFile = /hg19/out76_hg19.gem

chrFiles = FREEC_Linux64/chromosomes/

[sample]

mateFile=mpileup.txt
#mateCopyNumberFile=mpileup.txt_sample.cpn

inputFormat = pileup
mateOrientation = FR

[control]

[BAF]

SNPfile = hg19_snp131.SingleDiNucl.1based.txt
minimalCoveragePerPosition = 5


[target]
captureRegions=TruSeq_exome_targeted_regions.hg19.bed

[valeu]

If it looks like:

hs1 1631000 1731000 3.953692
hs1 1731000 1831000 3.953692
hs1 1831000 1931000 3.953692
hs1 1931000 2031000 3.953692

Circos will undestand it. You should add to the Circos config something like:

<plot>
show    = yes
type    = scatter
file = yourFile.CNP.circos.txt

glyph = rectangle
 glyph_size = 4
 fill_color = dgrey
 stroke_color = dgrey
 stroke_thickness = 1

min   = 0
max   = 6
#r0    = 0.7r
#r1    = 0.975r
r0    = 0.76r
r1    = 0.975r

background       = no
background_color = vvlgrey
background_stroke_color = black
background_stroke_thickness = 1

axis           = yes
axis_color     = lgrey
axis_thickness = 1
axis_spacing   = 1

<rules>

<rule>
importance   = 100
condition    = var(value) >= 6
value=6
glyph = rectangle
 glyph_size = 4
 fill_color = red
 stroke_color = red
 stroke_thickness = 1
</rule>

<rule>
importance   = 100
condition    = var(value) > 2.5 && var(value) < 6
glyph = rectangle
 glyph_size = 4
 fill_color = orange
 stroke_color = orange
 stroke_thickness = 1
</rule>

<rule>
importance   = 85
condition    = var(value) < 1.5
 glyph_size = 4
 fill_color = blue
 stroke_color = blue
 stroke_thickness = 1
</rule>

<rule>
show=0
importance   = 85
condition    = var(value) < 0
color = blue
</rule>
</rules>

</plot>

[dmacmillan]

Oh I forgot to ask about the circos format script. I plugged in my *ratios.txt file and got some output. But it does not look like circos format to me, how am I supposed to use the output given from this script?

[dmacmillan]

No I currently do not, to save time. I will try it with the GEM file though. I'll let you know how it goes!

[valeu]

do you use option "gemMappabilityFile" + "minMappabilityPerWindow"? usually, this helps since centromeric regions are not uniquely mappable. See http://bioinfo-out.curie.fr/projects...al.html#CONFIG

[dmacmillan]

Yes this helps. I was also wondering if there was a way to smooth out centromere calls? It seems to call very high or very short CNVs nearing centromeres.

[valeu]

Hi!

to avoid "point" CNVs you can use "minCNAlength=4" or more.. By default, it is 1.

Does it help?

[dmacmillan]

Hi, so I have been using this software for multiple analyses. However there are some quirks.

Here is a template of my config file:

[general]

chrLenFile = /projects/copy_num_ana/x07_controlfreec/hg18/res_all_genome.len
coefficientOfVariation = 0.062
ploidy = 2
outputDir = /projects/copy_num_ana/apollo_freec/4.24/RG/RG014/output
chrFiles = /projects/copy_num_ana/x07_controlfreec/hg18_fastas
forceGCcontentNormalization = 2

[sample]

mateFile = /projects/DLBCL/CNV/RG014/tumour/A01414_10_lanes_dupsFlagged.bam
inputFormat = BAM
mateOrientation = FR

[control]

mateFile = /projects/DLBCL/CNV/RG014/normal/A01443_9_lanes_dupsFlagged.bam
inputFormat = BAM
mateOrientation = FR

I am getting strange copy number calls, almost point-like gains in the middle of neutral or otherwise regions. Here is an example of what I am talking about (If you consider the red to be a gain, the blue to be neutral, and the green to be a loss):




Notice how the areas seem to correspond to areas with high GC-content and low reads. The IGV diagram above corresponds to chromosome 17 specifically the point right above RG034 at 3.98e10bp and 2.75 ratio, depicted below:


What is causing this? Has it been encountered before? And is there a way to correct this?

[valeu]

You are right, indeed, I changed the order of columns. It should be 2, 4, 10, 7, 8 and 5.

Sorry, I should have mentioned it.

[dmacmillan]

I managed to find out what file you were using, it turns out it is under tables > your assembly > all tracks > SNP130 (if your using hg18, for hg19 its 131) > hg18.snp130OrthoPt2Pa2Rm2

You then have to pick the columns you want which are:
chrom, chromstart, humanObserved, humanAllele, humanStrand

Even if you do this the file is not the same size as the one provided on the controlfreec website, also the columns are in a slightly different order.

[aggp11]

Hello valeu,

This is a very nice tool. Just the fact that it can still analyze the CNVs without needing a control sample. However, in your opinion do you think this could be used for certain Custom capture experiments? If so would it be better to instead of chromosome lengths provide the lengths of the targeted regions (within chromosomes).

Thanks,
Praful

[dmacmillan]

Could you be more specific? I have looked at the tables and gone through a lot of SNP's but none of them match the formatting of the file you include on the tutorial page where it gives the description of the BAF parameters. Can you go through step by step?

[valeu]

I download it through "Tables".

[dmacmillan]

Could you point out where you found it? When I look at hg18 downloads all I see are fasta files corresponding to SNPs.

[valeu]

I downloaded it from UCSC. It should be their standard format.