OneMap for genetic mapping of outbred crosses - anyone?

[MetteHHH]

Hello all!

This is my introductory post - I was in here a couple of years ago, but couldn't find my login info and changed my email since, so I re-registered.

Anyway, I would like to get to the point rather than small talk - so I decided to post here, rather than in the introduction section! :-)

I am using the R software package OneMap for genetic mapping of a pseudo-testcross. I discover my markers by SNP calling on a reduced representation library made with RAD-tagged Illumina output. What I would really love, would be to discover other people in a similar situation.

Is anyone here doing linkage mapping? Does anyone use OneMap? (The forum search didn't find anyone mentioning it, but I'm still hoping you might be out there...

[Nico55]

im in the same boat

I just started reading the one map manual today. How do you find it? I am very interested in seeing how it works for my populations.
Have you tried MST map? It works OK but it takes a lot of manual filtering of your SNPs to make them fit the cross models accepted by these programs. I guess this is how it has been done for years.
Sorry I’m no help right off the bat, but please let me know more about your experience with this. I will keep you posted on mine.

[Nico55]

package tkrplot

i can not get tkrplot to work

any ideas?

[MetteHHH]

Hi Nico55,

I am not familiar with MST map. I had a quick look at the documentation pages, but apparently it doesn't handle outbred data? That is what makes OneMap particularly useful for my purpose. What kind of data are you analysing?

I like the OneMap manual - it comes with a nice little tutorial dataset, and handles basic functions quite well. There is a bit of a lack of additional resources though - the help files in R is really all there is. I found the main author on Mendeley though, Gabriel Margarido, and he has been very helpful in answering questions.

What is tkrplot? A function in MST map?

I hope we can share some experiences as we go along with our respective projects. Thanks for answering the thread.

[lexa]

you can also take a look on the R package qtl. it can be used to create linkage maps.

[MetteHHH]

Thanks lexa, yes, OneMap is in fact a sort of add-on development of Rqtl. It is also R-based, but is able to handle more types of crosses than Rqtl does.

[lexa]

Hi MetteHHH, thanks for this info. I also tried OneMap but I found one think missing. if I'm not wrong, it cannot handle genomes like not A and not B, right? R/qtl can deal with this genotypes while creating a linkage map.

[MetteHHH]

Hi lexa,

Not sure I understood what you mean by "not A and not B" - do you mean null alleles? OneMap can handle those. In fact, I _think_ it can handle all types of markable markers, but they need to be recoded (which is a bit of a hassle) according to this table from the manual:



I had a colleague help me write a script for the recoding of the genotypes.

[lexa]

Hi lexa,

Not sure I understood what you mean by "not A and not B" - do you mean null alleles? OneMap can handle those. In fact, I _think_ it can handle all types of markable markers, but they need to be recoded (which is a bit of a hassle) according to this table from the manual:

thanks for this hint. now, I remember of recoding my data according to the table. right, it was a bit of a hassle.

[Nico55]

Correct, MTS map is not designed to work with out-crossers but you can separate your markers into groups that by their segregation behavior. Then run them in separate batches; DH and RIL, then join the maps. I have been doing this with some success but it is a pain this is why I’m so interested in onemap.
tkrmap is a dependency for onemap, it is working fine now, it was a tk/tcl header file issue.
I feel like onemap is SUPER slow. I’m running it on 64bit 40core with 170G memory but it is not using of these resources, any ideas on memory allocation, or threading?
The preliminary results are very promising.

[MetteHHH]

Nico55: I think I might have a useful tip for making onemap faster. I found that loading the massive twopts file into R slowed everything down tremendously. After creating seperate twopts files for each of my linkage groups, I can load just one at a time, and everything runs much faster. I hope that might help?

I have a question about picking the right criteria for marker inclusion: Onemap was originally made for microsats, where you get relatively limited marker sets. I have SNP markers from a reduced representation genomic sequencing, so even with conservative filtering I get marker sets of 8-10,000 markers. Since they are also from an outcross, a lot of them are of a not-too-informative sort (aaxab), and this gives me a rather high risk of false positives compared to the kind of data onemap handled originally. If I use the function order.seq (automated compare-try map construction) and look at the "safe" subset of markers (unambigously placed), I can identify some potentially false positives from the LOD/rf matrix. If I then force these markers off the map with the drop.marker function and reestimate the map, I generally get a slightly larger subset of "safe" markers. However, if I "wiggle" the number of markers in the initial "compare" fase of map construction, the actual markers in the safely placed subset varies quite a lot. Example:

One linkage group of 231 markers run with order.seq at LOD threshold 4 with 5 initial markers in the compare step placed 29 markers on the "safe" map. I dropped the four most dubious looking ones and reestimated the map - 38 markers were now included on the safe map. Reestimating with n.init of 6 and 8 gave similar sized sets, but each run results in a _different_ set of markers (of which many are not present on the other "safe" maps).

I realize that this can be expected, since the try function doesn't explore all positions, but only positions given by the previously placed markers. Placing one *relatively uninformative( aaxab marker will weaken the ability to place another one unambigously in a nearby position. But my question is: Is there a way to automate a procedure that "catches" a greater number of my 231 markers in this linkage group without lowering my LOD threshold?

One option would be to use "force" to place _every_ marker on the map, and then drop the bad ones, but doing this one-by-one would mean re-estimating the map hundreds of times(!) and with LGs with up to 450 markers this doesn't seem feasible.

I would rather have fewer markers with very high order credibility, but placing only about 10% of the markers in my LGs seems rather low...

Any input/ideas/discussion is welcome!

[catfisher]

input file problem

Hi MetteHHH and Nico55,

It's excited to get this threads that you guys communicated on. I also tried to use OneMap, but got the errors"Invalid marker codification" when importing data. Though I have carefully compared my file with the example file, I did not figure out the reason. Since you guys are more experienced with this. Here I attached a few markers. WOuld you please have a look at it? Thanks in advance!

[catfisher]

Hi，
I encountered the same issues you metioned about tk header file. How did you fix it then?

Correct, MTS map is not designed to work with out-crossers but you can separate your markers into groups that by their segregation behavior. Then run them in separate batches; DH and RIL, then join the maps. I have been doing this with some success but it is a pain this is why I’m so interested in onemap.
tkrmap is a dependency for onemap, it is working fine now, it was a tk/tcl header file issue.
I feel like onemap is SUPER slow. I’m running it on 64bit 40core with 170G memory but it is not using of these resources, any ideas on memory allocation, or threading?
The preliminary results are very promising.

[crop_genomics]

At present, I also want use onemap for F1 genetic map construction, and download the software, it showns nedd the tkrplot to work, but I can not use the tkrplot work

[Jeremy]

Hi MetteHHH and Nico55,

It's excited to get this threads that you guys communicated on. I also tried to use OneMap, but got the errors"Invalid marker codification" when importing data. Though I have carefully compared my file with the example file, I did not figure out the reason. Since you guys are more experienced with this. Here I attached a few markers. WOuld you please have a look at it? Thanks in advance!

This means one of your markers has a call that isn't allowed, for example you have a call like 'ba' instead of 'ab'.

The partial file contains correct calls, so the problem lies somewhere further in.