Reading the data into Haplin

Haplin reads data in two formats:

1. Haplin's own text file format;

2. PED format.

Both types of data are read in through the use of genDataRead function:

my.gen.data <- genDataRead( file.in = "my_gen_data.ped", file.out =
  "my_saved_gen_data", dir.out = ".", format = "ped" )
my.gen.data.haplin <- genDataRead( file.in = "my_gen_data_hap.dat",
  file.out = "my_saved_gen_data_hap", dir.out = ".", format = "haplin", n.vars = 0 )

The function reads in all the data in the file, creates ff objects to store the genetic information and data.frame to store covariate data (if any). These objects are saved in .RData and .ffData files, which can be later on easily uploaded to R (with genDataLoad) and re-used.

CAUTION: This can take a long time for large datasets (such as from GWAS analysis, e.g., reading in a 7 GB file will take ca.15 minutes), however, this needs to be run only once and then, the next time you need to use the data, use the genDataLoad function (see below). Be careful NOT TO DELETE the output files!

The genDataRead function returns a list object with two elements:

• cov.data - a data.frame with covariate data (if available in the input file);
• gen.data - the genetic part of data.

The above function reads in also any additional covariate data the user has through the parameter cov.file.in:

my.gen.data.haplin2 <- genDataRead( file.in = "my_gen_data_hap.dat",
  file.out = "my_saved_gen_data_hap2", cov.file.in = "add_cov.dat",
  dir.out = ".", format = "haplin" )

The file with the additional information should have a header with names of the data columns!

To see all the available arguments and usage examples, type:

?genDataRead

(this works also with any other function)

Preparing your data

After loading the data, it is necessary to pre-process it to the internal format used by haplin. This is done by evoking the command:

my.prepared.gen.data <- genDataPreprocess( data.in = my.gen.data, map.file =
  "my_gen_data.map", design = "triad", file.out = "my_prepared_gen_data",
  dir.out = "." )

CAUTION: This action can be very time-consuming for large datasets (e.g., estimated time for ca.45,000 SNPs and 1,600 individuals, a PED file of ca.700MB, is ca.6 minutes on a 7-core CPU). However, this needs to be done only once and the output, stored in small files, can be used for the subsequent analysis repeatedly. (See also section Choosing a subset of data

This will also create .RData and .ffData files, which take much less space than the input PED files. Be careful not to delete these files, as they can be re-used by simply loading into R (the genDataLoad function) right before Haplin analysis.

NOTE: The information on the my.prepared.gen.data object can be displayed by simply writing the name of the object.

Choosing a subset of data {#subset}

If you know that you want to focus your analysis on a certain region of the entire SNP set, or perhaps you're impatient and want to check out Haplin without waiting a long time for the preprocessing to finish, you can easily choose a subset of the data to pre-process and analyze. This can be done with the command:

gen.data.subset <- genDataGetPart( data.in = my.gen.data, markers = c( 3:15,22 ),
  file.out = "my_gen_data_subset", dir.out = "." )

This function allows you to specify the subset in various ways:

• markers - numeric vector with numbers indicating which markers to choose;
• indiv.ids - numeric vector giving IDs of individuals. CAUTION: in a standard PED file, individual IDs are not unique, so this will select all individuals with given numbers;
• rows - numeric vector giving the positions - this will select only these rows;
• cc - one or more values to choose based on the case-control status ('cc' column);
• sex - one or more values to choose based on the 'sex' column;
• … - if any additional covariate data are available in data.in, the user can choose based on values of these.

If you give a combination of these parameters, the result will be the intersection of the subsets defined by each of the parameters alone. The subset is then available in the gen.data.subset object and written to file.out file(s). These can be loaded and re-used multiple times.

Re-using the data

IMPORTANT: Remember that each time you start a new R session, you need to load the data into the memory with the command:

my.prepared.gen.data <- genDataLoad( filename = "my_prepared_gen_data",
  dir.in = "." )

This takes much less time than re-reading and running the data preparations! The output of the genDataPreprocess function (or genDataLoad) can then be used to run the analysis.