public List <MappedItemGroup> ReadFromFile(string fileName)
{
var result = new List <MappedItemGroup>();
XElement root = XElement.Load(fileName);
//Console.WriteLine("read locations ...");
Dictionary <string, SAMAlignedLocation> qmmap = root.ToSAMAlignedItems().ToSAMAlignedLocationMap();
//Console.WriteLine("read mapped items ...");
foreach (XElement groupEle in root.Element("subjectResult").Elements("subjectGroup"))
{
var group = new MappedItemGroup();
result.Add(group);
foreach (XElement mirnaEle in groupEle.Elements("subject"))
{
var mirna = new MappedItem();
group.Add(mirna);
mirna.Name = mirnaEle.Attribute("name").Value;
foreach (XElement regionEle in mirnaEle.Elements("region"))
{
var region = new SequenceRegionMapped();
mirna.MappedRegions.Add(region);
region.Region.Name = mirna.Name;
region.Region.ParseLocation(regionEle);
if (regionEle.Attribute("sequence") != null)
{
region.Region.Sequence = regionEle.Attribute("sequence").Value;
}
if (regionEle.Attribute("query_count_before_filter") != null)
{
region.QueryCountBeforeFilter = int.Parse(regionEle.Attribute("query_count_before_filter").Value);
}
if (regionEle.Attribute("pvalue") != null)
{
region.PValue = double.Parse(regionEle.Attribute("pvalue").Value);
}
foreach (XElement queryEle in regionEle.Elements("query"))
{
string qname = queryEle.Attribute("qname").Value;
string loc = queryEle.Attribute("loc").Value;
string key = SAMAlignedLocation.GetKey(qname, loc);
SAMAlignedLocation query = qmmap[key];
region.AlignedLocations.Add(query);
query.Features.Add(region.Region);
}
}
}
}
qmmap.Clear();
return(result);
}
public static void FilterMappedRegion(MappedItem item)
{
//deal with the item with multiple regions but one of them contains others
if (item.MappedRegions.Count > 1)
{
var removed = new List<SequenceRegionMapped>();
for (int i = 0; i < item.MappedRegions.Count; i++)
{
var regi = item.MappedRegions[i];
for (int j = i + 1; j < item.MappedRegions.Count; j++)
{
var regj = item.MappedRegions[j];
if (removed.Contains(regj))
{
continue;
}
var con = regi.Region.Contains(regj.Region);
if (con == 1)
{
//if i contains j and all mapped reads mapped to j, remove i
if (regi.AlignedLocations.All(m => regj.AlignedLocations.Contains(m)))
{
removed.Add(regi);
break;
}
//if i contains j and all mapped reads from j were contained in i, remove j
if (regj.AlignedLocations.All(m => regi.AlignedLocations.Contains(m)))
{
removed.Add(regj);
continue;
}
}
else if (con == -1)
{
//if j contains i and all mapped reads mapped to i, remove j
if (regj.AlignedLocations.All(m => regi.AlignedLocations.Contains(m)))
{
removed.Add(regj);
continue;
}
//if j contains i and all mapped reads from i were contained in j, remove i
if (regi.AlignedLocations.All(m => regj.AlignedLocations.Contains(m)))
{
removed.Add(regi);
break;
}
}
}
}
removed.ForEach(m => m.AlignedLocations.ForEach(l => l.Features.Remove(m.Region)));
item.MappedRegions.RemoveAll(m => removed.Contains(m));
}
}
public static void FilterMappedRegion(MappedItem item)
{
//deal with the item with multiple regions but one of them contains others
if (item.MappedRegions.Count > 1)
{
var removed = new List <SequenceRegionMapped>();
for (int i = 0; i < item.MappedRegions.Count; i++)
{
var regi = item.MappedRegions[i];
for (int j = i + 1; j < item.MappedRegions.Count; j++)
{
var regj = item.MappedRegions[j];
if (removed.Contains(regj))
{
continue;
}
var con = regi.Region.Contains(regj.Region);
if (con == 1)
{
//if i contains j and all mapped reads mapped to j, remove i
if (regi.AlignedLocations.All(m => regj.AlignedLocations.Contains(m)))
{
removed.Add(regi);
break;
}
//if i contains j and all mapped reads from j were contained in i, remove j
if (regj.AlignedLocations.All(m => regi.AlignedLocations.Contains(m)))
{
removed.Add(regj);
continue;
}
}
else if (con == -1)
{
//if j contains i and all mapped reads mapped to i, remove j
if (regj.AlignedLocations.All(m => regi.AlignedLocations.Contains(m)))
{
removed.Add(regj);
continue;
}
//if j contains i and all mapped reads from i were contained in j, remove i
if (regi.AlignedLocations.All(m => regj.AlignedLocations.Contains(m)))
{
removed.Add(regi);
break;
}
}
}
}
removed.ForEach(m => m.AlignedLocations.ForEach(l => l.Features.Remove(m.Region)));
item.MappedRegions.RemoveAll(m => removed.Contains(m));
}
}
public static List<MappedItem> GroupByName(IEnumerable<SequenceRegionMapped> regions)
{
var result = new List<MappedItem>();
foreach (var curregions in regions.GroupBy(m => m.Region.Name).ToList())
{
var mi = new MappedItem();
mi.Name = curregions.Key;
mi.MappedRegions.AddRange(curregions);
result.Add(mi);
}
return result;
}
public static List <MappedItem> GroupByName(IEnumerable <SequenceRegionMapped> regions)
{
var result = new List <MappedItem>();
foreach (var curregions in regions.GroupBy(m => m.Region.Name).ToList())
{
var mi = new MappedItem();
mi.Name = curregions.Key;
mi.MappedRegions.AddRange(curregions);
result.Add(mi);
}
return(result);
}
