How to map short reads to a distant genome?

ynwh · 08-02-2011, 08:11 AM

Hi,

My current project involves mapping short reads to the reference genome from another species (reads and reference genome belong to the same genus), e.g. mapping chimp reads to human genome. I want to know if this type of mapping is possible and if it is possible, Do the results has any meaning?

In other words, I want to know whether current short reads mapping algorithms (BWA, MAQ etc.) can give useful results when deal with somewhat distant genomes (same genus but different species)?

Many thanks

--Hao

GW_OK · 08-02-2011, 01:08 PM

I wouldn't call chimp and human that distant. I was once given some chimp labeled as human in a comedy of errors and it aligned just fine (using BWA). I thought I had found some REALLY rare mutations before the problem became obviously apparent, lol.

How are you defining useful and meaningful? Are you just wanting a scaffold to do some de novo assembly? This should work for that...

ynwh · 08-02-2011, 01:22 PM

Hi GW_OK,

Many thanks for your response. "Useful and meaningful" means when I find some PAV, CNV or mutations between two species, these findings are real and not artifacts of algorithm.

If chimp and human are not far enough, what do you think if I map two plants, e.g. wild rice to domesticated rice (O. sativa). Plant genomes are more dynamic then animals because of TEs

GW_OK · 08-02-2011, 02:11 PM

I would not use any CNV/SNP/Indel/mutation based on anything other than alignment of one species against itself. How could you tell if it was an alignment artifact or there really are differences between the species.

A disparate species alignment is really only good for de novo scaffolding, as far as I know (anyone else can feel free to correct me).

Edit: Or are you attempting to use this assembly as some sort of round-about BLAST?

ynwh · 08-02-2011, 05:41 PM

Yeah. What I want to do is to blast short reads against the reference genome. But the reads number are too high to do a BLAST search. Do you have any recommendation on what program can do such job?

gringer · 08-03-2011, 06:56 AM

Quote:

Yeah. What I want to do is to blast short reads against the reference genome. But the reads number are too high to do a BLAST search. Do you have any recommendation on what program can do such job?

Bowtie is fairly fast at its job, and is able to handle up to 3 SNPs in the seed (by default 29bp), but setting it for that low stringency will make it take much longer and produce more false positive hits.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Genome Res De novo bacterial genome sequencing: millions of very short reads assembly	b_seite	Literature Watch	1	10-05-2017 12:26 AM
How to map 454 reads/contigs to a mitochondrial genome?	fruktimport	Bioinformatics	2	03-28-2011 11:35 AM
Short reads fragments of genome...	hicham	Bioinformatics	2	03-24-2010 05:15 AM
looking for reference genome based assembler for short-reads	zchou	Bioinformatics	3	12-16-2009 09:13 PM
Haman genome alignment with short reads	ptongyoo	Bioinformatics	4	04-14-2009 06:27 PM

08-02-2011, 08:11 AM	#1
ynwh Member Location: MD Join Date: Jul 2011 Posts: 16	How to map short reads to a distant genome? Hi, My current project involves mapping short reads to the reference genome from another species (reads and reference genome belong to the same genus), e.g. mapping chimp reads to human genome. I want to know if this type of mapping is possible and if it is possible, Do the results has any meaning? In other words, I want to know whether current short reads mapping algorithms (BWA, MAQ etc.) can give useful results when deal with somewhat distant genomes (same genus but different species)? Many thanks --Hao

08-02-2011, 01:08 PM	#2
GW_OK Senior Member Location: Oklahoma Join Date: Sep 2009 Posts: 411	I wouldn't call chimp and human that distant. I was once given some chimp labeled as human in a comedy of errors and it aligned just fine (using BWA). I thought I had found some REALLY rare mutations before the problem became obviously apparent, lol. How are you defining useful and meaningful? Are you just wanting a scaffold to do some de novo assembly? This should work for that...

08-02-2011, 01:22 PM	#3
ynwh Member Location: MD Join Date: Jul 2011 Posts: 16	Hi GW_OK, Many thanks for your response. "Useful and meaningful" means when I find some PAV, CNV or mutations between two species, these findings are real and not artifacts of algorithm. If chimp and human are not far enough, what do you think if I map two plants, e.g. wild rice to domesticated rice (O. sativa). Plant genomes are more dynamic then animals because of TEs

08-02-2011, 02:11 PM	#4
GW_OK Senior Member Location: Oklahoma Join Date: Sep 2009 Posts: 411	I would not use any CNV/SNP/Indel/mutation based on anything other than alignment of one species against itself. How could you tell if it was an alignment artifact or there really are differences between the species. A disparate species alignment is really only good for de novo scaffolding, as far as I know (anyone else can feel free to correct me). Edit: Or are you attempting to use this assembly as some sort of round-about BLAST?

08-02-2011, 05:41 PM	#5
ynwh Member Location: MD Join Date: Jul 2011 Posts: 16	Yeah. What I want to do is to blast short reads against the reference genome. But the reads number are too high to do a BLAST search. Do you have any recommendation on what program can do such job?