public static List<SmallRNASequenceContig> GetTopContig(List<SmallRNASequence> sequences, double minimumOverlapRate, int maximumExtensionBase)
{
List<SmallRNASequenceContig> result = new List<SmallRNASequenceContig>();
foreach (var seq in sequences)
{
var contig = new SmallRNASequenceContig();
contig.ContigSequence = seq.Sequence;
contig.ContigCount = seq.Count;
contig.Sequences.Add(seq);
result.Add(contig);
}
return GetTopContig(result, minimumOverlapRate, maximumExtensionBase);
}
public static List<SmallRNASequenceContig> BuildContigByIdenticalSequence(Dictionary<string, List<SmallRNASequence>> counts, int topNumber = int.MaxValue)
{
//Get unique sequences
List<IGrouping<string, SmallRNASequence>> sequences;
if (topNumber == int.MaxValue)
{
sequences = (from map in counts.Values
let smap = map
from seq in smap
select seq).GroupBy(m => m.Sequence).
OrderByDescending(m => m.Sum(l => l.Count)).ToList();
}
else
{
sequences = (from map in counts.Values
let smap = map.Take(Math.Min(map.Count, topNumber)).ToList()
from seq in smap
select seq).GroupBy(m => m.Sequence).
OrderByDescending(m => m.Sum(l => l.Count)).ToList();
}
//Initialize Sequence~Contig map
var resultMap = new Dictionary<string, SmallRNASequenceContig>();
foreach (var seq in sequences)
{
var contig = new SmallRNASequenceContig();
contig.ContigSequence = seq.Key;
resultMap[seq.Key] = contig;
}
//Add smallRNAsequence into Sequence~Contig map
foreach (var map in counts.Values)
{
foreach (var count in map)
{
SmallRNASequenceContig contig;
if (resultMap.TryGetValue(count.Sequence, out contig))
{
contig.Sequences.Add(count);
}
}
}
//Initialize config count
foreach (var seq in resultMap.Values)
{
seq.ContigCount = seq.Sequences.Sum(l => l.Count);
}
return (from v in resultMap.Values
orderby v.ContigCount descending
select v).ToList();
}
public static List<SmallRNASequenceContig> BuildMiniContig(List<SmallRNASequenceContig> contigs, int topNumber)
{
var result = new List<SmallRNASequenceContig>();
for (int i = 0; i < contigs.Count && i < topNumber; i++)
{
var contig = contigs[i];
var seqCount = (from seq in contig.Sequences.GroupBy(l => l.Sequence)
select new { Seq = seq.Key, SeqCount = seq.Sum(l => l.Count), SeqIndex = contig.ContigSequence.IndexOf(seq.Key) }).OrderByDescending(l => l.SeqCount).ToList();
//seqCount.ForEach(l => Console.WriteLine("{0}\t{1}\t{2}", l.Seq, l.SeqCount, l.SeqIndex));
var fq = seqCount[0];
var list = seqCount.Where(l => Math.Abs(l.SeqIndex - fq.SeqIndex) < 3 && l.SeqCount >= fq.SeqCount * 0.1).ToList();
var start = list.Min(l => l.SeqIndex);
var end = list.Max(l => l.SeqIndex + l.Seq.Length);
var contigSeq = contig.ContigSequence.Substring(start, end - start);
var contigSequences = new HashSet<string>(list.ConvertAll(l => l.Seq));
var miniContig = new SmallRNASequenceContig();
miniContig.ContigSequence = contigSeq;
miniContig.ContigCount = list.Sum(l => l.SeqCount);
miniContig.Sequences.AddRange(contig.Sequences.Where(l => contigSequences.Contains(l.Sequence)));
result.Add(miniContig);
}
return result;
}