#55
EVALUATION OF AUTOMATED HIGH-RESOLUTION HLA TYPING SOFTWARE.
Erik H. Rozemuller PhD 1, Wietse Mulder PhD 1 and Marcel G.J. Tilanus PhD 1. 1 Genome Diagnostics, c/o UMCU, bldg H04.312, Utrecht, P.O.Box 85500, Netherlands .
In the recent years high resolution HLA typing has improved significantly, in part due to increased knowledge of the full-length allele sequences. Moreover, instruments and reagents, both accurate and affordable, came within reach of diagnostic laboratories. Nowadays every HLA typing centre can easily obtain high resolution typing data. However, it became more cumbersome to analyze the data set properly. This is even truer for SBT typing, where hundreds of data points are generated. Although the launch of various software packages helped to automate this process, still a numurous samples need manual reviewing due to anomalies in DNA sequence profiles.
We took up the gage to investigate the nature of failures in the analysis process. We reasoned that a better understanding would help to improve the underlying algorithm and to implement an expert system.
We collected DNA sequences of numerous samples that were a) derived from various origins, b) obtained using various SBT strategies, including commercial kits as well as home brewed reagents. In addition, sequences were included that originated from either homozygous or heterozygous samples or even after allele separation techniques like GSSP, GSAP. All sequencing files were imported into SBTengine
and analyzed consecutively.
We have analyzed the performance by monitoring to what extend the DNA sequences analysis led to the correct allele assignment. The majority of failures were due to bad quality sequences but in addition several structural sequence anomalies were observed.
We present a comprehensive overview of the data and the cause of the rejected samples. This study leads to several new insights that may be implemented into software algorithms and thereby facilitating accurate and affordable tissue typing.