I have some new information on the insertion sites of three Exelixis insertions. These were flagged by me as having insertion locations in FlyBase that were on a different chromosome than in the GDP database. The GDP localizations were based on the GDP trimming and alignment of flanking sequences determined by Exelixis. Exelixis sent Roger Hoskins phred sequence files of the untrimmed inverse PCR sequence reads. I have reanalyzed the untrimmed phred sequences for these three insertions and have also done limited genetic segregation analysis of the insertions.
1) c03484 (FBti0041099) - This is annotated in FlyBase as an allele of CG11030 on chromosome 2, presumably based on the gene assignment reported by Exelixis in Thibault et al. 2004 (FBrf0174227). FlyBase does not have a sequence location. FlyBase has a GenBank accession (CZ468676) for the 3' flank that was submitted by Roger Hoskins in 2005. I retrimmed the 3' flank and submitted an updated 48 bp 3' flank to GenBank. I also submitted a new 225 bp 5' flank to GenBank (KS718208). Both of the flanks are repetitive, although they do not align to the canonical sequences for any of the standard D. melanogaster TE families. When I blast the flanks, I get alignments with > 90% matches to sites on every chromosome arm. The best alignment of the merged 5' and 3' flanks to the release 6 assembly of the D. melanogaster reference genomic sequence is a 268/269 match to a site in the CG40470 gene on 3L. I ordered a stock carrying c03484 from the BDSC (#11281) with a genotype w1118; PBac{w+mC=PB}CG11030c03484/CyO. All of the flies in the stock have an orange eye color and curly wings. In December 2017, I crossed males from the stock X yw virgin females and found that the orange eye color segregated from the curly wings, indicating that the w+ transgene is on chromosome 2. My conclusion is that, although the PBac is inserted on chromosome 2, as reported by Exelixis, the flanking sequences are repetitive and there is no evidence supporting the CG11030 gene assignment or mapping it to any specific site on chromosome 2.
2) c04788 (FBti0041172) - This is annotated in FlyBase as an allele of Desi on chromosome arm 3R, presumably based on the gene assignment reported by Exelixis in Thibault et al. 2004 (FBrf0174227). FlyBase does not have a sequence location. FlyBase has GenBank accessions for both flanks. Both flanks begin at the vector-flank junction and both map to the same release 6 sequence location:  2R:6979155- in the vimar gene. I ordered a stock carrying c03484 from the BDSC (#11641) with a genotype w1118; PBac{w+mC=PB}Desic04788. In December 2017, I crossed males of this stock to virgin females with yw; Sp/CyO; Dr/TM3 (BDSC 59967). All of the progeny were w+, indicating that the transgene was homozygous in the original stock. I crossed yw; CyO; TM3 male progeny x yw virgin females and, separately w1118/yw; CyO; TM3 virgin female progeny x w1118 males. In both crosses, the w+ transgene marker segregated from CyO, indicating that the PBac insertion is on chromosome 2, consistent with the sequence location based on the flanking sequence alignment.
3) e03004 (FBti0041661) - This is annotated in FlyBase as an allele of CG14838 on chromosome arm 3L, presumably based on the gene assignment reported by Exelixis in Thibault et al. 2004 (FBrf0174227). FlyBase does not have a sequence location or GenBank accessions for either flank. However, the BDSC stock (18107) has a genotype w1118; l(2)**/CyO; PBac{w+mC=RB}CG14838e03004 and indeed, all of the flies in the stock have curly wings. I became suspicious of the Exelixis mapping of the insertion when I saw that the insertion site mapped by the GDP mapped to 2R in the Dark gene. Both of the flanks are short and truncated by the restriction site used for iPCR. I submitted both flanks to GenBank in October 2017. The 23 bp 3' flank (GenBank KS718228), based on two sequence runs giving the same sequence, aligns 23/23 to a unique site in the release 6 assembly  2R:17020497-. The 14 bp 5' flank (GenBank KS718229) aligns 14/14 to at least two sites, but one of those is the same site as the 23 bp 3' flank. I ordered the BDSC stock and found that all of the flies have a yellow eye color and curly wings. I crossed males of the stock x yw virgin females and found that the w+ transgene segregated from CyO. Kevin Cook found that e03004 is lethal in combination with Df(2R)BSC359, a deletion of the region of Dark. It also failed to complement the lethality of Dark82.In summary, there is evidence from the flanking sequences and genetic segregation that e03004 is inserted at release 6  2R:17020497- in the Dark gene and that it is a lethal allele of Dark.