Segmentation violation when Augustus reads in hints.gff

[katharina]

Originally posted by Robert Bruccoleri in the old forum on 12.04.2013 - 00:29

I am trying to use Augustus 2.7 to model genes in a novel fungal organism, and I'm getting a segmentation violation when the first hint is being read in.

We have a good assembled genome sequence, gene models produced with Cegma and Snap, and a nice dataset of RNAseq reads from Illumina. Using Tophat and Cufflinks, we get a number of putative transcripts consistent with Cegma and Snap.

I was able to make an initial Augustus species configuration using the Cegma results.

I'm following the tutorial in the documentation:

http://bioinf.uni-greifswald.de/bioinf/ ... seq.Tophat

but step6 is failing: Here's the command I'm running:

Code: Select all

augustus --species=fungusamongus --extrinsicCfgFile=extrinsic.cfg --alternatives-from-evidence=true --hintsfile=hints.gff --allow_hinted_splicesites=atac genome_seq.fa

I'm getting a segmentation violation when it reads in the first hint. Here's the first hint:

Code: Select all

scaffold00001 b2h intron 102395 102452 0 . . pri=4;src=E

Here's the traceback from gdb and some poking around with gdb from the core dump:

Code: Select all

(gdb) where
#0  0x0000000000485c0d in FeatureCollection::readGFFFile (
    this=0x7fff9a76a040, filename=0xabaab88 "hints.gff")
    at extrinsicinfo.cc:2201
#1  0x000000000040305d in main (argc=7, argv=0x7fff9a76aed8)
    at augustus.cc:154
Program terminated with signal 11, Segmentation fault.
#0  0x0000000000485c0d in FeatureCollection::readGFFFile (
    this=0x7fff9a76a040, filename=0xabaab88 "hints.gff")
    at extrinsicinfo.cc:2201
2201                    double gradequot = fti.gradequots[sourcenum][f.gradeclass];
(gdb) p f.gradeclass
$1 = 0
(gdb) p sourcenum
$2 = 2
(gdb) p fti.gradequots[sourcenum][f.gradeclass]
One of the arguments you tried to pass to operator[] could not be converted to what the function wants.
(gdb) l 2150,2210

2160    void FeatureCollection::readGFFFile(const char *filename){
2161        /*
2162         * Read in the configuration file for extrinsic features.
2163         */
2164        readExtrinsicCFGFile();
2165        int predictionStart, predictionEnd;
2166        try {
2167          predictionStart = Properties::getIntProperty( "predictionStart" ) - 1;
2168        } catch (...) {
2169          predictionStart = 0;
2170        }
2171        if (predictionStart < 0)
2172            predictionStart = 0;
2173        try {
2174          predictionEnd = Properties::getIntProperty( "predictionEnd" ) - 1;
2175        } catch (...) {
2176          predictionEnd = INT_MAX;
2177        }
2178
2179        try {
2180            datei.open(filename);
2181            if( !datei ) {
2182                cerr << "FeatureCollection::readGFFFile( " << filename << " ) : Could not open the file!!!"\
 << endl;
2183                throw ProjectError();
2184            }
2185
2186            /*
2187             * read in line by line
2188             *
2189             */
2190            Feature f;
2191            string seqname;
2192            datei >> comment >> ws;
2193            while (datei) {
2194                datei >> f >> comment >> ws;
2195                if (f.end >= predictionStart && f.start <= predictionEnd && f.type != -1) {
2196                    f.start -= predictionStart;
2197                    f.end -= predictionStart;
2198                    FeatureTypeInfo& fti = typeInfo[f.type];
2199                    int sourcenum = esource(f.esource);
2200                    f.gradeclass = fti.gradeclass(sourcenum, f.score);
2201                    double gradequot = fti.gradequots[sourcenum][f.gradeclass];
2202                    // set the general values if applicable
2203                    if (!(fti.bonus < 0)) { // general bonus/malus
2204                        f.bonus = fti.bonus * gradequot * BONUS_FACTOR;
2205                        f.malus = fti.malus;
2206                        // Let the intron bonus depend on the length
2207                        if (f.type == intronF) {
2208                            int laenge = f.end - f.start + 1;
2209                            double intronGeoProb = IntronModel::getGeoProb(); // 1-1/1730 (mal)
2210                            double igenicGeoProb = IGenicModel::getGeoProb(); // 0.9999;

I ran the program with the debugger, and it died on the very first record.
Can anyone provide advice on what I'm going wrong or what to try next?
Thanks.
Robert Bruccoleri
bruc@acm.org

[katharina]

Originally posted by Robert Bruccoleri in the old forum on 12.04.2013 - 01:28

I just learned that if I reference the sample extrinsic.cfg file augustus.2.7/config/extrinsic/extrinsic.E.cfg instead of extrinsic.cfg file in the tutorial:

Code: Select all

[SOURCES] 
M RM E W
exonpart 1 .992 M 1 1e+100 RM 1 1 E 1 1 W 1 1.005 
intron 1 .34 M 1 1e+100 RM 1 1 E 1 1e5 W 1 1 
CDSpart 1 1 0.985 M 1 1e+100 RM 1 1 E 1 1 W 1 1 
UTRpart 1 1 0.985 M 1 1e+100 RM 1 1 E 1 1 W 1 1 
nonexonpart 1 1 M 1 1e+100 RM 1 1.01 E 1 1 W 1 1

the segmentation violation does not occur. I guess I didn't understand what to do about creating this file.