Quality control in AFND

Each population in AFND is named according to the combination of the country name, geographical location and ethnic group when available, to describe the population in as much detail as possible.

In order to identify the polymorphic region studied in a given population an additional word is incorporated at the end of the name, except for HLA populations which were the first populations entered in the database.

If another set of individuals from a given population, which was geographically and ethnically similar to an existing population in the database, was submitted, a consecutive number was assigned to that population to differentiate the two populations.

In some populations, individuals are living or born in a different country from their original ethnic background (i.e., immigrants from a different country). In these cases, the name of the original ethnic background was included in the name of the population, and the country was defined as the current location in which individuals were living/born. For these populations, users can search data using either country as filter.

Finally, many countries may have populations typed from different sources. Thus, if users are interested in populations from one country, we recommend that the search is performed for populations from that country initially, then users can filter populations according to specific sources. E.g. Anthropology.

In AFND, we have organised the populations by geographical regions. In collaboration with the IDAWG and HLA-NET consortiums, we have defined 12 geographical regions.

Allele names may have been changed in the IMGT/HLA database (Official Nomenclature Database) after their original submission for different reasons. For example, the sequence of the allele A*01:34N was shown to be expressed at low levels and the allele was renamed to A*01:01:38L in March 2011. Thus, in AFND, we have inputted the allele frequency under the new allele code.

Some populations were typed only by serology. In these instances, we have converted data into the IMGT/HLA database nomenclature.

AFND collects data at different level of resolution. We automatically generate all possible low resolution allele names based on the IMGT/HLA catalogue. For example, the allele A*01:01:01:01 is automatically split to generate frequencies for A*01, A*01:01 and A*01:01:01 alleles.

AFND receives quarterly reports from the IMGT/HLA, i.e. every new release. We update AFND according to the latest release immediately after this notification.

Based on the high diversity of HLA alleles, one may expect that the frequencies of alleles should not exceed 50%. However, we have detected some populations which allele frequencies are over 50%. This may be the case of certain loci that have a low number of alleles, such as DPA1, DQB1, etc., or in some populations that have very few alleles at a given locus. We have also examined those populations that have > 75% of the individuals with that allele.

In some cases, authors have excluded in their publications those alleles reported at low frequencies. To specify this, we have included a sentence in the demographics of the population.

For some populations all haplotypes are not necessarily listed. Sometimes this is because not all haplotypes are listed in the publication because they are at a very low frequency. As far as possible, we have listed all haplotypes greater than 1%. When the population is large, we have added haplotypes at lower percentages.

We decided to capture both allele frequencies at high and low resolution, by summing high resolution data to produce low resolution frequency data. For example, the Northern Ireland population in AFND will have frequencies for A*25:01 = 0.0200, A*25:02 = 0.0010 and the sum of these two A*25 = 0.0210.

If it is known that an allele/gene has been tested for and not found, we have entered 0.000. If the allele/gene has not been tested for, the frequency column is left in blank.

In some cases, the number individuals sampled (typed) for each locus within the same population are different. Thus, users are advised to check demographic data of the pops they have searched for to ascertain possible deviations in data. In instances, where alleles could not be distinguished, we would put the frequency under the first allele but adding a note in demographic information.

AFND receives reports from the IPD-KIR every new release. We update AFND according to the latest release immediately after this notification.