I have had to grapple with this too for a bit until I discovered a quick way to convert a phased VCF file to the rehh .hap and .map standard input without the need to include the recode_allele flag in the tool.

- First, I used PLINK2 to convert the phased VCF to SHAPEIT2 format .haps + .lengend

-Then I used sed and awk to make some few changes in the files to produce the rehh input file formats

Here is the code

--------------------------------------------------------------------------------------------------------

plink2 \

--chr 11 \

--export hapslegend \

--vcf phased.vcf.gz \

--out chr11

sed '1d' chr11.legend | awk '{print $1"\t""11""\t"$2"\t"$4"\t"$3}' > chr11.map

sed 's/0/2/g' chr11.haps > chr11.hap

--------------------------------------------------------------------------------------------------------

I used the --chr flag because I was interested in chr 11

Key points from the code

- sed '1d' : deletes the header line from the legend file (id position a0 a1)

- awk '{print $1"\t""11""\t"$2"\t"$4"\t"$3}' : prints the content of the legend file, addind chr number between id and position, while switching the a0 and a1 columns to rehh format.

NB: a0 is usually the minor allele and a1is usually the major allele (https://www.cog-genomics.org/plink/2.0/formats#haps)

This produces the .map file

- sed 's/0/2/g' : The SHAPEIT .haps file is encoded 0/1, 0 corresponding to a0 and 1 corresponding to a1 in the legend file. rehh encodes 1 as ancestral allele which would correspond to 1 (major usually ref or ancestral allele in the .haps file). So no need to change this. Hence we are only left with 0 (minor allele usually alt or derived) in the .haps file to convert to 2 (rehh enconding for alt allele).

https://mathgen.stats.ox.ac.uk/genetics_software/shapeit/shapeit.html#haplegsample

I ascertained my ref alleles by comparing with the 1000 Genomes Phase 3 dataset

Note that ascertainment of ancestral/derived is more complicated. However, one can limit sites to those with no ambiguity in allele frequency and ref/alt status. For example some minor alleles usually turn out to be ref alleles.

i feel a little bit confused by the jargon you used, a little bit of elaboration would be nice, but i think you mean how to load genome data files into R  and analyzing/handling them using rehh package;

it would be something like below to load files you mentioned:

hap

Dear Hassan,

Thanks for the reply and sorry for being unclear.

Already rehh and gplot package are installed on my system and they work well with example files provided by rehh (mentioned in tutorial).

I have a list of SNPs and to be able to use rehh, I need two input file:

1)".hap" file: standard haplotype format

2)".map" file for SNPs

As you may know almost all of haplotype files on databases are in.ped or VCF (in the 1000 genomes project case) format. Right now I am seeking for a way to obtain .hap file of SNPs list which I have. Really I come across nothing!! 

Could you please suggest a tools which can make .hap file?

there is a tool to convert from VCF to other formats {you have to compile/build it yourself i think}

http://vcftools.sourceforge.net

but for a worst case scenario i suggest you could write a program to convert *.vcf files to *.hap files, it seems vcf files have all the information you need to build hap files out of them; another thing is to try other R packages that work with 1000genomes file formats(if one exists);

thanks for your suggestions. They are big deal for me!!

Dear Hamid,

I have similar problem with you. But I am concerning to use fastphase input file instead of standard hap file. Right now, I am trying to get map input file as map.inp, actually on how to get ancestral and derived allele column on that map file. I will follow your discussion since it was very helpfull too for me... Thank you...

Hi everyone

I have a very similar problem too. I have output files of SHAPEIT (.haps and .sample) and I need to convert them in one of the two input formats compatible with rehh package. If someone has any suggestion I will appreciate. Thanks

I suggest a collective effort by us to consult seasoned statistical bioinformaticians who have employed these tools in seminal papers. I think this impinges on the important need for reproducibility in modern day genomic research. I have come across a series papers in this regard.  Please do find attached. There are also some key gurus  in this field at the London School of Hygiene and tropical medicine and the Wellcome trust Sanger Institute, UK. Honestly, I think the best solution is to initiate a formal collaboration with the bioinformaticians and the statistical geneticists. Goodluck

Hi everyone!

If you have any question regarding input files for REHH2.0 R package, fill free to address directly to me. While working on my projects, I have wrote a lot of scripts to convert 1000 Genome and other data for REHH package. I will explain what exactly should be done with your original data and how to avoid possible error messages during iHS, XP-EHH calculations with the use of REHH2.0 package. Just send me a personal message via this site. To add, I'm open for collaborative projects in bioinformatics.  

Here is the script I wrote for BEAGLE to REHH2.0 R package conversion.

https://gist.github.com/Gennady-Khvorykh/89ac0ff3dac1c1c23d1b313f99fafb0a

First you need to convert your plink files to fastPHASE files using fcgene program. Then you can use fastPHASE to phase the genotypes. After that you can easily import the genotypes into rehh and conduct the requires tests. Now, a challenging bit is the map file in rehh which assumes that you know the ancestary allele.Usually for livestock I had difficulties findinv this info. However, not all the tests in rehh require knowledge of the ancestary alleles so choose the ones that dont require it or use the fcgene SNP info file at your own risk. Note that you can actually force the major allele to be written as allele1 in the fcgene snp info file if you wish (please read fcgene manual to know how).I really hope this helps!

Let me develop the ideas provided by Mohammed Ali Al Abri.

1. fastPHASE input files can be generated directly by plink. Just use `--recode fastphase-1chr ` option. One input file should include one chromosome. For example,

plink --file chr22 --recode fastphase-1chr --out chr22

2. Ancestry alleles are required for iHS metric and are not required for cross population one, e.g., XP-EHH.

3. fastPHASE is out of date. It was fast indeed a few years ago. Now it is slowPHASE :) I advice to use SHAPEIT or BEAGLE.

Great. I would try these ideas and revert

I have had to grapple with this too for a bit until I discovered a quick way to convert a phased VCF file to the rehh .hap and .map standard input without the need to include the recode_allele flag in the tool.

- First, I used PLINK2 to convert the phased VCF to SHAPEIT2 format .haps + .lengend

-Then I used sed and awk to make some few changes in the files to produce the rehh input file formats

Here is the code

--------------------------------------------------------------------------------------------------------

plink2 \

--chr 11 \

--export hapslegend \

--vcf phased.vcf.gz \

--out chr11

sed '1d' chr11.legend | awk '{print $1"\t""11""\t"$2"\t"$4"\t"$3}' > chr11.map

sed 's/0/2/g' chr11.haps > chr11.hap

--------------------------------------------------------------------------------------------------------

I used the --chr flag because I was interested in chr 11

Key points from the code

- sed '1d' : deletes the header line from the legend file (id position a0 a1)

- awk '{print $1"\t""11""\t"$2"\t"$4"\t"$3}' : prints the content of the legend file, addind chr number between id and position, while switching the a0 and a1 columns to rehh format.

NB: a0 is usually the minor allele and a1is usually the major allele (https://www.cog-genomics.org/plink/2.0/formats#haps)

This produces the .map file

- sed 's/0/2/g' : The SHAPEIT .haps file is encoded 0/1, 0 corresponding to a0 and 1 corresponding to a1 in the legend file. rehh encodes 1 as ancestral allele which would correspond to 1 (major usually ref or ancestral allele in the .haps file). So no need to change this. Hence we are only left with 0 (minor allele usually alt or derived) in the .haps file to convert to 2 (rehh enconding for alt allele).

https://mathgen.stats.ox.ac.uk/genetics_software/shapeit/shapeit.html#haplegsample

I ascertained my ref alleles by comparing with the 1000 Genomes Phase 3 dataset

Note that ascertainment of ancestral/derived is more complicated. However, one can limit sites to those with no ambiguity in allele frequency and ref/alt status. For example some minor alleles usually turn out to be ref alleles.

Gennady Khvorykh, I faced a problem. I convert beagle to rehh using REHH2.0 R package and I got ; *.hap, and *.inp files. Then I use R package, rehh, to calculate iHS, XPEHH, and Rsb but the analysis failed when I run the following script " hap

This mos likely due duplicate by position in your vcf

Endashaw Terefe, from your message it is not clear how you got REHH2.0 input files (*.inp, *.hap). What exactly have you done to get them?

Just in case, I convert BEAGLE output files (*.bgl, *.markers) into REHH2.0 input files (*.inp, *.hap) with custom R script https://gist.github.com/Gennady-Khvorykh/89ac0ff3dac1c1c23d1b313f99fafb0a . Feel free to test it.

Gennady Khvorykh thank you for the response. You are correct. I convert *.bgl, *.markers to *.inp, *.hap using the REHH2.0. Then I tried to calculate the iHS, XPEHH and Rsb using the *.inp and *.hap files. At this time I got this message

...

"Map info: 325918 markers declared for chromosome 1 .

Haplotype input file in standard format assumed.

Error: The number of markers in the haplotype file (1900709) is not equal

to the number of markers in the map file (325918).

Conversion stopped." ...

I will try the script and let you know the result

Hi Endashaw Terefe

I see that your map fiile has the .inp extension. Could you try changing the recode.allele = FALSE to recode.allele = TRUE?

However, if you use the recent version of BEAGLE (BEAGLE 5.0: http://faculty.washington.edu/browning/beagle/beagle.12Jul19.0df.jar) to phase your genotypes and get a VCF file, then you may find these linux-based scripts helpful.

https://github.com/esohkevin/popgen/blob/master/phased_vcf2rehh.sh

https://github.com/esohkevin/popgen/blob/master/helpmsg.sh

You have to get both scripts into the same path so that you get help messages with the conversion.

The script can convert your phased VCF genotypes to the standard rehh .hap + .map input so that you can use the recode.allele = FALSE option.

Hello,

Can anyone help me with importing files into rehh that are in a simple dataframe format? I have a table of genotypes already coded as 0=missing/ 1=ref/ 2=non-ref, and a table of SNP data. Simplified version of the data:

### Generating example data

## Creating example SNP map

V1

Hello William Hamilton ,

I took a look at your question and noticed what you were not getting right.

- First, REHH does not allow multi-chromosome files. Your data must be split by chromosomes

- REHH expects you to pass it connections and not the data already loaded. REHH loads the data given the connections.

So I worked on these points and got the following:

#--- Convert the variables into the appropriate types

V1