Category 8: genes which have been deliberately separated from their CG's
because there was more than one such CG and we had no time to sort them
out earlier.  Here, then, the requirement is to decide which CG NOT to
merge with the gene.  I have only made a preliminary analysis.
These should be easy by Clustals of proteins.
>Ag5r FBgn0015010 = CG9538|FBan0009538|CT26984| 
>And FBgn0011273 = CG17769|FBan0017769|CT39380|FBan0017769  
DO NOT MERGE
(And FBgn0011273 does not align with CG17770|FBan0017770|CT39382|.
CG5024|CT16147 is NOT And by alignment, but all 3 seem to be 
Calmodulin-related)
>Cnx99A FBgn0015622 = CG11958|FBan0011958|CT4036|
(CG9906|FBan0009906|CT27896| most likely Cnx14D; it's clearly related to
but not the same as Cnx99A)
>drl FBgn0015380  = CG10758|FBan0010758|CT30136| AND 
CG17348|FBan0017348|CT30099|
MERGE
>Fur1 FBgn0004509 = CG10772|FBan0010772|CT29938|
(CG17049 has no homology to known fly genes (by BLASTN and BLASTP at NCBI))
>fz  FBgn0001085 581 aa = CG3646|FBan0003646|CT12089|329aa (fz aa 1-329) AND 
CG17697|FBan0017697|CT39176| 166aa (fz aa 416-581)
MERGE
(aa 330-415 still missing! fz spans over 90 kb, 60 kb gap OK, some of the
missing aas are probably in the gap)
>lds  FBgn0002542 974aa = CG2684|FBan0002684|CT9083| 1061aa
DO NOT MERGE
(CG10445|FBan0010445|CT29324 is not lds, see Clustal;
Aubrey, the two CG's don't seem to be overlapping in the GenBank annotation,
do you have additional info?)
>Lk6  FBgn0017581 = CG6929|FBan0006929|CT21460| AND CG17342|FBan0017342|CT21464|
MERGE
>Mgstl  FBgn0025814 = CG1742|FBan0001742|CT5060|
Mgstl must be CG1742: CG12628/CT35150 is on wrong arm
(CG12628 is probably a related protein)
>Pdp1 FBgn0016694 = CG17888|FBan0017888|CT39809|   
(Rab-RP4 FBgn0015794 =CG3129|FBan0003129|CT10657|)
>ras FBgn0003204 = CG11485|FBan0011485|CT36321| AND CG1799|FBan0001799|CT5398|
MERGE
>RpS14a FBgn0004403 = CG1524|FBan0001524|CT31809|
(RpS14b FBgn0004404 = CG1527|FBan0001527|CT3937|)
>RpS15A
??
'RpS15A must be CG12324: CT6271/CT5068 CG2033 wrong arm'
CONFLICT: both are good matches to RpS15A. By nucleotide alignment 
RpS15A cDNAs  align better to where CG2033 is on scaffold. Could cytology
have been derived from location of CG12324? See Clustal.
>ScpX FBgn0015808 = CG17320|FBan0017320|CT29258|CT29260  best match 
DO NOT MERGE
(CG17597 looks like it's a adjacent related gene, but both ScpX cDNAs 
in GB match  CG17320 better by nucleotide alignment)
>sif FBgn0019652 = CG5256|FBan0005256|CT16789| AND CG5406|FBan0005406|CT17116
MERGE
>Taf80 FBgn0010356 = CG7704|FBan0007704|CT23475|F
DO NOT MERGE
(CG6751 also annotated as Taf80 probably because of nucleotide sequence 
overlap, its protein does not align with GB Acc. U06460 for Taf80) 
>term FBgn0003683=CG4216|FBan0004216|CT13826|
DO NOT MERGE,SUSPECT
(term matches CG4216 throughout its length (428) but matches only the
first 261 aa of CG7271 which is on the other strand. Adjacent 
related gene?)
>anon-68Ed FBgn0017424 = CG11538|FBan0011538|CT17864|
> BcDNA:GH07188  FBgn0027578 = CG14526|FBan0014526|CT34253|
DO NOT MERGE
(By nucleotide alignment CG14526 is  BcDNA:GH07188 ; this is probably
a multigene family)
> BcDNA:LD21735  834 = CG18617|FBan0018617|CT22945| AND             
          CG7679|FBan0007679|CT24090 |CT30413
MERGE
> BG:DS00929.8  FBgn0028922 = CG4182|FBan0004182|CT12287|
(l(2)35Bg FBgn0001977 = CG4180|FBan0004180|CT13516|)
>JTBR FBgn0025820 152 aa = CG1935|FBan0001935|CT5993| 152 aa
>dlt FBgn0024246  871 aa = CG12021|FBan0012021|CT1747| 871 aa
DO NOT MERGE
(from Category 1 list)
>l(2)dtl FBgn0013548 = CG11295|FBan0011295|CT31521|
DO NOT MERGE
l(2)tid(FBgn0002174) = CG5504|FBan0005504|CT17450|FBan0005504
(note: CG5504 is erroneously annotated as l(2)dtl in public view of GadFly; 
CG11295 is correctly annotated as l(2)dtl)
(from Category 1 list)
>Lnk FBgn0028717 = CG5908|FBan0005908|CT18543| AND CG17367|FBan0017367|CT38370|
MERGE
>Pp4-19C = CG1596|FBan0001596|CT3517 AND CG1459|FBan0001459|CT3537|
MERGE
>RpL8 FBgn0024939 = CG1263|FBan0001263|CT2579|CT37695
DO NOT MERGE
CG18334|FBan0018334|CT37687 is NOT RpL8!!!
Uch FBgn0010288 = CG4265|FBan0004265|CT10917| 
rost FBgn0011705 = CG9552|FBan0009552|CT27002|
(CG9555 is adjacent, related, do not merge.)
Rbp9 = CG3151|FBan0003151|CT10570|CT38165| 
(CT10570 extends 229 aa beyond Rbp9 from CDS set, check when annotating.)
(CG5808 is not Rbp9.)
fw FBgn0001083 = CG1500|FBan0001500|CT3745|
(CG1500 has more than 150 aa upstream of the start of fw in CDS set,
and is missing some of the C terminus of the fw in CDS set. 
CG9095 has some identity with fw but is not fw.)
α-Adaptin FBgn0015567 = CG4260|FBan0004260|CT13966|
(CG9113 is not α-Adaptin.)
Acp33A FBgn0023414 = CG15841|FBan0015841|CT40022| AND CG6555|FBan0006555|CT20399|
(MERGE: Acp33A has 2 possible ORFs on one mRNA, these are the 2 ORFs.)
nAcRβ-64B FBgn0000038 = CG12606|FBan0012606|CT34874| AND CG11348|FBan0011348|CT31654|
(MERGE; CG11348 and CG12606 start at the same ATG but CG11348 doesn't splice out
the first intron and CDS truncates within the intron)
Mst84Db FBgn0004173 = CG17934|FBan0017934|CT39960|
(There happens to be an ORF on the opposite strand: CG14599. Do not merge.
The translation of CG17934 matches that of Mst84Db annotated in Acc. X67703) 
I think there is a bit of a mixup between LIP1 and Lip1,
they both have the same accession (Y14366). That accession 
belongs to Lip1 (Lipase 1), I don't think there is an accession 
associated with LIP1 (LAR-interacting protein 1).
...
Lip1 FBgn0023496=CG7279|FBan0007279|CT22413|FBan0007279
(CG7279 is now wrongly annotated as LIP1 FBgn0024747)
HDAC3 FBgn0025825 438 aa = CG2128|FBan0002128|CT6896|438 aa
(Aubrey: Protein ID is missing from gene report: AAC83649.1 
This was missing from the CDS set)
mit(1)15 FBgn0004643 = CG9900|FBan0009900|CT25082|