Augustus output - how is the incompatible hint groups determined?

[orionzhou]

Hi, I have a question about the augustus (v3.0.2, v3.0.1) output - specifically when it reports the number of "hint groups fully obeyed" and "incompatible hint groups".

I used a test genomic sequence (~3kb) which has a typical two-exon gene, the sequence is available here.

I know exactly where the exons begin and end, so I put these information in a hint file (I also pasted it below since it's really straightforward to understand).

scaffold_0      .       start   101     103     10      .       .       pri=2;grp=start;src=M
scaffold_0      .       CDSpart 101     379     10      .       .       pri=2;grp=cds1;src=M
scaffold_0      .       CDSpart 3040    3210    10      .       .       pri=2;grp=cds2;src=M
scaffold_0      .       intronpart      380     3039    10      .       .       pri=2;grp=intron;src=M
scaffold_0      .       stop    3208    3210    10      .       .       pri=2;grp=stop;src=M

Then I ran augustus:

augustus --species=arabidopsis --hintsfile=hints.gff --gff3=on test.fas

which gives the following output:

##gff-version 3
# This output was generated with AUGUSTUS (version 3.0.2).
# ----- prediction on sequence number 1 (length = 3232, name = scaffold_0) -----
#
# Predicted genes for sequence number 1 on both strands
# start gene g1
scaffold_0    AUGUSTUS    gene    76    3232    0.08    +    .    ID=g1
scaffold_0    AUGUSTUS    transcript    76    3232    0.08    +    .    ID=g1.t1;Parent=g1
scaffold_0    AUGUSTUS    transcription_start_site    76    76    .    +    .    Parent=g1.t1
scaffold_0    AUGUSTUS    exon    76    379    .    +    .    Parent=g1.t1
scaffold_0    AUGUSTUS    start_codon    101    103    .    +    0    Parent=g1.t1
scaffold_0    AUGUSTUS    intron    380    3039    1    +    .    Parent=g1.t1
scaffold_0    AUGUSTUS    CDS    101    379    1    +    0    ID=g1.t1.cds;Parent=g1.t1
scaffold_0    AUGUSTUS    CDS    3040    3210    1    +    0    ID=g1.t1.cds;Parent=g1.t1
scaffold_0    AUGUSTUS    exon    3040    3232    .    +    .    Parent=g1.t1
scaffold_0    AUGUSTUS    stop_codon    3208    3210    .    +    0    Parent=g1.t1
scaffold_0    AUGUSTUS    transcription_end_site    3232    3232    .    +    .    Parent=g1.t1
......

# Evidence for and against this transcript:
# % of transcript supported by hints (any source): 60
# CDS exons: 2/2
#      M:   2
# CDS introns: 1/1
#      M:   1
# 5'UTR exons and introns: 0/1
# 3'UTR exons and introns: 0/1
# hint groups fully obeyed: 1
#      M:   1 (intron)
# incompatible hint groups: 4
#      M:   4 (start,cds1,cds2,stop)
# end gene g1
###

As you can see augustus is predicting exactly the same gene structure as the hint file. However, it is claiming that it is following only 1 hint (intron) and "incompatible" with all other hints (start codon, stop codon, plus two CDS). How could this be possible?

PS: I changed "CDSpart" to "CDS" and augustus does seem to obey the rule - but what if I don't know the exact boundary of a CDS and have to put in "CDSpart"?

PPS: I have tried augustus v3.0.2, v3.0.1, v2.7.1 and problem all exists, however, v2.5.5 gave the right summary report though.

Any comments would be greatly appreciated!