public override IEnumerable <string> Process()
{
Progress.SetMessage("reading fasta file ...");
var faMap = SequenceUtils.Read(new FastaFormat(), options.FastaFile).ToDictionary(m => m.Name);
Progress.SetMessage("{0} sequences read ...", faMap.Count);
using (StreamWriter sw = new StreamWriter(options.OutputFile))
{
Progress.SetMessage("reading gff file ...");
var gffs = GtfItemFile.ReadFromFile(options.GffFile);
}
return(new string[] { options.OutputFile });
}
public GtfDistanceExporter(string gtfFile, string gtfKey = "exon")
{
Console.WriteLine("reading gtf file " + gtfFile + " ...");
this.maps = CollectionUtils.ToGroupDictionary(GtfItemFile.ReadFromFile(gtfFile, gtfKey), m => m.Seqname.StringAfter("chr"));
Console.WriteLine("reading gtf file " + gtfFile + " done");
this.header = string.Format("distance_{0}\tdistance_{0}_position\tdistance_gene\tdistance_in_gene", gtfKey);
this.emptyStr = new String('\t', header.Count(m => m == '\t'));
//sort the gtf items by locus
foreach (var lst in maps.Values)
{
lst.Sort((m1, m2) => m1.Start.CompareTo(m2.Start));
}
}
/// <summary>
/// Get 1-based coordinate from file. Bed format will be automatically translated.
/// </summary>
/// <param name="coordinateFile">source file</param>
/// <param name="gtfFeature">if it's gtf format, which feature name will be used as gene_id</param>
/// <param name="bedAsGtf">if bed already be 1-based</param>
/// <returns></returns>
public static List <GtfItem> GetSequenceRegions(string coordinateFile, string gtfFeature = "", bool bedAsGtf = false)
{
bool isBedFormat = IsBedFormat(coordinateFile);
List <GtfItem> result;
if (isBedFormat)
{
//bed is zero-based, and the end is not included in the sequence region
//https://genome.ucsc.edu/FAQ/FAQformat.html#format1
//gtf is 1-based, and the end is included in the sequence region
//http://useast.ensembl.org/info/website/upload/gff.html
//since pos in sam format is 1-based, we need to convert beditem to gtfitem.
//http://samtools.sourceforge.net/SAMv1.pdf
var bedItems = new BedItemFile <BedItem>().ReadFromFile(coordinateFile);
if (!bedAsGtf)
{
bedItems.ForEach(m => m.Start++);
}
result = bedItems.ConvertAll(m => new GtfItem(m));
}
else
{
result = GtfItemFile.ReadFromFile(coordinateFile).ToList();
if (!string.IsNullOrEmpty(gtfFeature))
{
result.RemoveAll(m => !m.Feature.Equals(gtfFeature));
}
result.ForEach(m =>
{
if (m.Attributes.Contains("gene_id \""))
{
m.GeneId = m.Attributes.StringAfter("gene_id \"").StringBefore("\"");
}
else if (m.Attributes.Contains("ID="))
{
m.GeneId = m.Attributes.StringAfter("ID=").StringBefore(";");
}
if (m.Attributes.Contains("gene_name \""))
{
m.Name = m.Attributes.StringAfter("gene_name \"").StringBefore("\"");
}
else if (m.Attributes.Contains("Name="))
{
m.Name = m.Attributes.StringAfter("Name=").StringBefore(";");
}
if (string.IsNullOrEmpty(m.GeneId) && !string.IsNullOrEmpty(m.Name))
{
m.GeneId = m.Name;
}
if (!string.IsNullOrEmpty(m.GeneId) && string.IsNullOrEmpty(m.Name))
{
m.Name = m.GeneId;
}
if (string.IsNullOrEmpty(m.GeneId))
{
m.GeneId = m.Attributes;
m.Name = m.Attributes;
}
});
}
return(result);
}