comparing results from two different reference genomes

BAJ · 02-23-2009, 01:12 AM

we have a Solexa experiment that seems to be contaminated with a different genome than the one we were originally aiming at. The genomes have very different sizes (mouse vs. pombe) and if I understand correctly the quality scores from the fastq output correctly they are dependent on the size of the reference genome.
In our particular example it seems to me that the smaller genome will always get lower scores (due to the smaller reference genome). Is there a way to account for that and make the quality scores comparable?

To clarify a bit my confusion:
I got a Gerald output in s_1_sequence.txt with a reference genome to pombe that starts like this:
@PF2:1:1:1644:1100
ATGAATTTCAGCCTCTGGTCAGGCAGGGTTCCTTTT
+PF2:1:1:1644:1100
OOOOOOOOPOKOOPPOKKOOOOKOOEKKOOGGGGGA
@PF2:1:1:1702:1050
ACCAAGCGCAAATTTACGATTTAATTAGTATTTATA
+PF2:1:1:1702:1050
OPOOOOPOOOOOOOOOOOOOOOOOOOOHOOCHAHHE
@PF2:1:1:1532:1901
TTCAAATATTCCTGATCCAATGACAAGTTGAACCGT

And I get another file for a mouse genome as reference:
@PF2:1:1:1644:1100
ATGAATTTCAGCCTCTGGTCAGGCAGGGTTCCTTTT
+PF2:1:1:1644:1100
VVVVVVVVVVOVVVVVOOVVVVMVVCMNVVQQRRRE
@PF2:1:1:1702:1050
ACCAAGCGCAAATTTACGATTTAATTAGTATTTATA
+PF2:1:1:1702:1050
VVVVVVVVVVVVVVVVVVVVVVVVVVVIVVGRCRRP
@PF2:1:1:1532:1901
TTCAAATATTCCTGATCCAATGACAAGTTGAACCGT

clearly the quality scores are different.
Which makes me believe that not only peak information but also alignment information is used. Peak information is used because I see for the same sequence different quality scores. There reference genome is used for the calculation of the quality score because for the exact same clusters different scores are being obtained with different reference genomes.
Now the questions: How do I make the values comparable for different reference genomes. I want to identify sequences that align better to one reference genome compared to another one in order to get some understanding about the possible contamination.

Any comment is appreciated.

Thanks,

Bernd

BAJ · 02-24-2009, 12:25 AM

I just heard back from techsupport at illumina that this is a property of pipeline 1.0 whereas pipeline 1.3.2 is independent of the alignment.

bioinfosm · 02-24-2009, 08:38 AM

Ohh, so thats changed in 1.3
Yes this was the case as Illumina updated its quality values based on Gerald alignments. If you check the quality values in Bustard folder, they should match irrespective of reference

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
comparing results by cuffdiff, edgeR, DESeq	PFS	Bioinformatics	5	03-12-2014 04:01 AM
Convert 1000-Genomes-proje BAM to FASTA (aligned to reference, grouped by chromosome)	ce.log	Bioinformatics	17	01-14-2014 12:35 AM
Where can I find FASTQ files along with reference genomes for various species?	gvivek	Bioinformatics	1	09-09-2011 04:30 AM
Reference-guided cDNA assembly using related genomes	ohofmann	Bioinformatics	0	02-08-2011 10:20 AM
Key Reference Whole Genomes now available for NextGENe® Software	SoftGenetics	Vendor Forum	0	08-25-2009 07:14 AM

02-23-2009, 01:12 AM	#1
BAJ Member Location: Paris Join Date: Nov 2008 Posts: 15	comparing results from two different reference genomes we have a Solexa experiment that seems to be contaminated with a different genome than the one we were originally aiming at. The genomes have very different sizes (mouse vs. pombe) and if I understand correctly the quality scores from the fastq output correctly they are dependent on the size of the reference genome. In our particular example it seems to me that the smaller genome will always get lower scores (due to the smaller reference genome). Is there a way to account for that and make the quality scores comparable? To clarify a bit my confusion: I got a Gerald output in s_1_sequence.txt with a reference genome to pombe that starts like this: @PF2:1:1:1644:1100 ATGAATTTCAGCCTCTGGTCAGGCAGGGTTCCTTTT +PF2:1:1:1644:1100 OOOOOOOOPOKOOPPOKKOOOOKOOEKKOOGGGGGA @PF2:1:1:1702:1050 ACCAAGCGCAAATTTACGATTTAATTAGTATTTATA +PF2:1:1:1702:1050 OPOOOOPOOOOOOOOOOOOOOOOOOOOHOOCHAHHE @PF2:1:1:1532:1901 TTCAAATATTCCTGATCCAATGACAAGTTGAACCGT And I get another file for a mouse genome as reference: @PF2:1:1:1644:1100 ATGAATTTCAGCCTCTGGTCAGGCAGGGTTCCTTTT +PF2:1:1:1644:1100 VVVVVVVVVVOVVVVVOOVVVVMVVCMNVVQQRRRE @PF2:1:1:1702:1050 ACCAAGCGCAAATTTACGATTTAATTAGTATTTATA +PF2:1:1:1702:1050 VVVVVVVVVVVVVVVVVVVVVVVVVVVIVVGRCRRP @PF2:1:1:1532:1901 TTCAAATATTCCTGATCCAATGACAAGTTGAACCGT clearly the quality scores are different. Which makes me believe that not only peak information but also alignment information is used. Peak information is used because I see for the same sequence different quality scores. There reference genome is used for the calculation of the quality score because for the exact same clusters different scores are being obtained with different reference genomes. Now the questions: How do I make the values comparable for different reference genomes. I want to identify sequences that align better to one reference genome compared to another one in order to get some understanding about the possible contamination. Any comment is appreciated. Thanks, Bernd Last edited by BAJ; 02-23-2009 at 08:52 AM.

02-24-2009, 12:25 AM	#2
BAJ Member Location: Paris Join Date: Nov 2008 Posts: 15	I just heard back from techsupport at illumina that this is a property of pipeline 1.0 whereas pipeline 1.3.2 is independent of the alignment.

02-24-2009, 08:38 AM	#3
bioinfosm Senior Member Location: USA Join Date: Jan 2008 Posts: 482	Ohh, so thats changed in 1.3 Yes this was the case as Illumina updated its quality values based on Gerald alignments. If you check the quality values in Bustard folder, they should match irrespective of reference