I'm trying to run cuffmerge using samples from cufflinks output files for these.

I did

cuffmerge assemblies.txt

where assmblies.txt had

./A/transcripts.gtf
./B/transcripts.gtf

If you look at the cufflinks output isoforms for these two you get:

 tracking_id     class_code      nearest_ref_id  gene_id gene_short_name tss_id  locus   length  coverage        FPKM    FPKM_conf_lo    FPKM_conf_hi    FPKM_status
CUFF.1.1        -       -       CUFF.1  -       -       ERCC-00003:3-1058       1055    401.337 16651.1 15842.2 17460   OK
CUFF.2.1        -       -       CUFF.2  -       -       ERCC-00004:0-597        597     2845.28 118048  114864  121231  OK
CUFF.3.1        -       -       CUFF.3  -       -       ERCC-00009:0-999        999     267.03  11078.8 10396   11761.6 OK
CUFF.4.1        -       -       CUFF.4  -       -       ERCC-00019:4-600        596     19.7057 817.567 552.313 1082.82 OK
CUFF.5.1        -       -       CUFF.5  -       -       ERCC-00022:6-751        745     77.8041 3228.01 2780.37 3675.66 OK
CUFF.6.1        -       -       CUFF.6  -       -       ERCC-00025:3-1948       1945    7.95538 330.06  250.591 409.529 OK
CUFF.7.1        -       -       CUFF.7  -       -       ERCC-00034:8-980        972     9.71542 403.082 270.55  535.615 OK
CUFF.8.1        -       -       CUFF.8  -       -       ERCC-00035:2-1069       1067    30.8168 1278.56 1055.99 1501.12 OK
CUFF.9.1        -       -       CUFF.9  -       -       ERCC-00042:1-1045       1044    198.638 8241.3  7668.49 8814.11 OK

and

tracking_id     class_code      nearest_ref_id  gene_id gene_short_name tss_id  locus   length  coverage        FPKM    FPKM_conf_lo    FPKM_conf_hi    FPKM_status
CUFF.1.1        -       -       CUFF.1  -       -       ERCC-00003:3-1018       1015    478.931 17910.6 17093.9 18727.2 OK
CUFF.2.1        -       -       CUFF.2  -       -       ERCC-00004:0-598        598     3221.58 120477  117424  123531  OK
CUFF.3.1        -       -       CUFF.3  -       -       ERCC-00009:3-1005       1002    298.562 11165.3 10515.2 11815.4 OK
CUFF.4.1        -       -       CUFF.4  -       -       ERCC-00019:26-614       588     16.4694 615.906 394.666 837.145 OK
CUFF.5.1        -       -       CUFF.5  -       -       ERCC-00022:4-743        739     99.3447 3715.2  3256.15 4174.25 OK
CUFF.6.1        -       -       CUFF.6  -       -       ERCC-00025:470-1946     1476    10.2811 384.482 289.104 479.861 OK
CUFF.7.1        -       -       CUFF.7  -       -       ERCC-00034:10-980       970     8.96558 335.286 220.284 450.288 OK
CUFF.8.1        -       -       CUFF.8  -       -       ERCC-00035:12-1099      1087    28.0996 1050.84 861.34  1240.34 OK
CUFF.9.1        -       -       CUFF.9  -       -       ERCC-00042:0-1015       1015    217.56  8136.1  7585.68 8686.51 OK

but if you look at the cuffmerge output isoforms.fpkm_tracking you get :

tracking_id     class_code      nearest_ref_id  gene_id gene_short_name tss_id  locus   length  coverage        FPKM    FPKM_conf_lo    FPKM_conf_hi    FPKM_status
CUFF.1.1        -       -       CUFF.1  -       -       ERCC-00002:1-1132       1131    0.000108688     1       0       0       OK
CUFF.2.1        -       -       CUFF.2  -       -       ERCC-00003:3-1058       1055    0.000100395     1       0       0       OK
CUFF.3.1        -       -       CUFF.3  -       -       ERCC-00004:0-598        598     5.79575e-05     1       0       0       OK
CUFF.4.1        -       -       CUFF.4  -       -       ERCC-00009:0-1005       1005    9.70485e-05     1       0       0       OK
CUFF.5.1        -       -       CUFF.5  -       -       ERCC-00019:4-614        610     5.7424e-05      1       0       0       OK
CUFF.6.1        -       -       CUFF.6  -       -       ERCC-00022:4-751        747     7.1974e-05      1       0       0       OK
CUFF.7.1        -       -       CUFF.7  -       -       ERCC-00025:3-1948       1945    0.000165918     1       0       0       OK

Notice how it gives you 1 for all the FPKM values and it gives you a really small coverage. I was under the impression the merged isoforms would have similar FPKM like taking average of the original samples...

Anyone know how to fix this?