w molecular markers, se quences similar to GenBank deposited sequences were filtered out to avoid identification of already known SSR and SNP sequences, especially the ones previously iden tified by turbot. Pilot microarray platform A custom 2 x 105 K array was printed with turbot se quences from the Turbot 3 database by Agilent Technologies. In order to study the orientation of the non annotated sequences and their possible gene expression, false annotation of genes and identify possible NATs, oligos were designed in both orientations, forward and reverse. Oligo design was done by using Repeat Masker to eliminate low complexity regions, and then OligoArray 2. 1 software to do the design itself. Cross hybridization between oligos was checked by BLAST searches against the entire Turbot 3 database and oligos with 3 putative cross hybridizations were re moved.

A total number of 96,292 oligos were printed and almost half of the array contained oligos also designed with the opposite orientation. This pilot micro array also included all default positive and negative con trols defined by the company. Microarray hybridization The same samples of immune tissues used for library construction and Sanger sequencing and those from the brain pituitary gonad axis used for 454 sequencing were used for hybridization with the pilot micro array. A total of four microarrays were used, two for the reproductive system and two for the immune system. Hybridizations were performed at the Universidad de Santiago de Compostela Functional Genomics Platform by the Agilent Technology Gene Expression Unit using a 1 colour labeling protocol.

This method demonstrated very similar Drug_discovery performances to the 2 colour protocol. Briefly, 50 ng of total RNA were labelled using the Low Input Quick Amp Labeling Kit, One Color. cRNA was prepared for overnight hybridization with the corresponding buffers during 17 h at 65 C and washed on the following day. Hybridized slides were scanned using an Agilent G2565B microarray scanner. Pilot microarray data processing, filtration, and identification of NATs The hybridization signal was captured and processed using an Agilent scanner. The scanner images were segmented with the Agilent Feature Extraction Software using protocol GE1 v5 95. Extended dynamic range implemented in the Agilent software was applied to avoid saturation in the highest intensity range.

Agilent feature extraction pro duced the raw data for further pre processing. The processed signal value was chosen as statistical for the absolute hybridization signal. The filtration process was made in two steps. First, the features which did not conform with any of the following well established quality criteria were filtered, non uniform pixel distributed outliers and population repli cate outliers according to the default Agilent feature extraction criteria, features whose ratio between pro cessed signal and their error was below 2, spots not differentiated from background signal, features b