RAD-Seq to construct genetic maps


question Questions
  • How to analyze RAD sequencing data for a genetic map study?
objectives Objectives
  • SNP calling from RAD sequencing data
  • Find and correct haplotypes
  • Create input files for genetic map building software
requirements Requirements

time Time estimation: 8 hours


This tutorial is based on the analysis described in publication. Further information about the pipeline is available from the official STACKS website. The authors developed a genetic map in the spotted gar and presented data from a single linkage group. The gar genetic map is an F1 pseudotest cross between two parents and 94 of their F1 progeny. They took the markers that appeared in one of the linkage groups and worked backwards to provide the raw reads from all of the stacks contributing to that linkage group.

This tutorial re-analyzes these data through to genotype determination. These data do not require demultiplexing and do not need processing though Process Radtags tool.


In this tutorial, we will deal with:

  1. Pretreatments
    1. Data upload
  2. Building loci using STACKS
  3. Genotypes determination


Data upload

The original data is available at STACKS website and the subset used here is findable on Zenodo.

hands_on Hands-on: Data upload

  1. Create a new history for this RAD-seq exercise.
  2. Import Fasta files from parents and 20 progeny.

    comment Comments

    If you are using the GenOuest Galaxy instance, you can load the dataset using ‘Shared Data’ -> ‘Data Libraries’ -> ‘1 Galaxy teaching folder’ -> ‘EnginesOn’ -> ‘RADseq’ -> ‘Genetic map’

    • Copy the link location
    • Open the Galaxy Upload Manager
    • Select Paste/Fetch Data
    • Paste the following links into the text field
    • Press Start

    As default, Galaxy takes the link as name. It does not link the dataset to a database or a reference genome.

Building loci using STACKS

Run Stacks: De novo map Galaxy tool. This program will run ustacks, cstacks, and sstacks on each individual, accounting for the alignments of each read.

comment Comment

Information on denovo_map.pl and its parameters can be found online: https://creskolab.uoregon.edu/stacks/comp/denovo_map.php.

hands_on Hands-On: Stacks: De novo map

Stacks: De novo map tool: Run Stacks selecting the Genetic map usage. Specify each parent as a sample in the appropriate box, then each of the 20 progeny and specify a CP Cross type, 3 for the Minimum number of identical raw reads required to create a stack, 3 for minimum number of identical raw reads required to create a stack in ‘progeny’ individuals, 3 for the number of mismatches allowed between loci when building the catalog and activate the option “remove, or break up, highly repetitive RAD-Tags in the ustacks program”.

De novo map input

Once Stacks has completed running, you will see 5 new data collections and 8 datasets.

The output of de novo map

Investigate the output files: result.log and catalog.* (snps, alleles and tags).

Looking at the first file, denovo_map.log, you can see the command line used and the start as end execution time.

De novo map log file

Then are the different STACKS steps:


De novo map:ustacks log


De novo map: cstacks

question Question

  1. Can you identify the meanning of the number 425?
  2. Looking at the catalog.tags file, identify specific and shared loci from each individual. Count the number of catalog loci coming from the first individual, from the second, and find on both parents.

solution Solution

  1. Here, the catalog is made with 459 tags, 425 coming from the “reference individual”, a female. Some of these 425 can be shared with the other parent.
  2. 3500


De novo map: sstacks log

Lastly, genotypes is executed. It searches for markers identified on the parents and the associate progenies’ haplotypes. If the first parent have a GA (ex: aatggtgtGgtccctcgtAc) and AC (ex: aatggtgtAgtccctcgtCc) haplotypes, and the second parent only a GA (ex: aatggtgtGgtccctcgtAc) haplotype, STACKS declares an ab/aa marker for this locus. Genotypes program then associate GA to a and AC to b and then scan progeny to determine which haplotype is found on each of them.

De novo map: genotypes log

De novo map: genotypes log

De novo map: genotypes log

Finally, 447 loci, markers, are kept to generate the batch_1.genotypess_1.tsv file. 459 loci are stored on the observed haplotype file batch_1.haplotypes_1.tsv.

Matches files

Here are sample1.snps (left) and sample2.snps (right)

De novo map matches files

Catalog_ID (= catalog Stacks_ID) is composed by the Stack_ID from the “reference” individual, here sample1, but number is different from sample2 Stack_ID. Thus, in the catalog.alleles.tsv, the Stack_ID number 3 corresponds to the Stack_ID number 16 from sample2!

You can inspect the matches files (you maybe have to change the tsv datatype to a tabular one to correctly display the datasets).

Male and female matches files

Consider catalog SNPs 27 & 28, on the 302 catalog locus:

De novo map matches considering catalog SNPs

nd the corresponding catalog haplotypes, 3 on the 4 possible (AA, AT, GT but no GA):

De novo map matches considering catalog haplotypes

heterozygosity is observed on each parent (one ab, the other ac) and there are 19 genotypes for the 22 individuals.

De novo map macthes: markers

We can then see that Stack_ID 330 for female corresponds to the 39 for male:

De novo map matches: male and female

Genotypes determination

hands_on Hands-on: Stacks: Genotypes

Stacks: genotypes tool: Re-Run the last step of Stacks: De novo map pipeline specifying more options as:

  1. The genetic map type (ie F1, F2 (left figure, F1xF1), Double Haploid, Back Cross (F1xF0), Cross Pollination (right figure, F1 or F2 but resulting from the cross of pure homozygous parents))

    The genetic map type F2 The genetic map CrossPollination

  2. Genotyping options output file type for input in genetic mapper tools (ie JoinMap, R/qtl, …). Observe that the R/qtl format for an F2 cross type can be an input for MapMaker or Carthagene.

  3. Thresholds concerning a minimal number of progeny and/or minimum stacks depth to consider a locus

  4. Make Automated Corrections to the Data. This option allows the user to have the program automatically correct some types of errors. This setting can correct errors with the homozygous tags verification in the progeny by confirming the presence or absence of the SNP. If SNP detection model can’t identify a site as heterygous or homozygous, that site is temporarily tagged as homozygous to facilitate the search, by sstacks, in concordance with the loci catalog. If a second allele is detected on the catalog (ie, in parents) and is found on a progeny with a weak frequency (<10% of a stack reads number), the genotypes program can correct the genotype. Additionally, it will delete a homozygous genotype on a particular individual if locus genotype is supported by less than 5 reads. Corrected genotypes are marked uppercase.

Here is an example of a locus originally marked as homozygous before automatic correction because an allele is supported by less than 5 reads. After correction, this locus is marked as heterozygous.

Automatic correction of genotypes

You can re-run Stacks: genotypes tool: modifying the number of genotyped progeny to consider a marker and thus be more or less stringent. Compare results.

Genotypes.tsv files

One line by locus, one column by individual (aa, ab, AB if automatic correction applied, bb, bc, …) with observed genotype for each locus:

Genotypes.tsv file overview

Genotypes.txt files

One line by individual, and for each individual, for each catalog locus, genotype:

Genotypes.txt file overview

Haplotypes.tsv files

One line by locus, one column by individual (aa, ab, AB if automatic correction applied, bb, bc, …) with observed genotype for each locus:

Haplotypes.tsv file overview

question Question

  1. The use of the deleverage algorithm allows to not consider loci obtained from merging more than 3 stacks. Why 3 if biologically, you are waiting something related to 2 for diploid organisms?
  2. Re-execute Stacks: De novo map pipeline modifying the p-value treshold for the SNP model. What is the difference regarding to unverified haplotypes ?

solution Solution

  1. This value of 3 is important to use if we don’t want to blacklist loci for whom 99.9% of individuals have one and/or the alt allele and 0.01% have a third one, resulting of a sequencing error.
  2. We see a moficiation of the number of unverified haplotypes


In this tutorial, we have analyzed real RAD sequencing data to extract useful information, such as genotypes and haplotypes to generate input files for downstream genetic map creation. This approach can be summarized with the following scheme:

The genetic map tutorial workflow

Useful literature

Further information, including links to documentation and original publications, regarding the tools, analysis techniques and the interpretation of results described in this tutorial can be found here.

congratulations Congratulations on successfully completing this tutorial!