//EInfo, Get Databases
//EGQuery, Search all databases and return database/count
//ESearch, Search a database and return a list of UIDs
//ESummary, Turn UID into Document Summary
//EFetch, Turn UID into formatted data records
//EFetch, see http://www.ncbi.nlm.nih.gov/books/NBK25499/table/chapter4.chapter4_table1/?report=objectonly
//ELink, List links, and more
//ELink, see http://www.ncbi.nlm.nih.gov/books/NBK25499/#chapter4.ELink
//EPost, Append UIDs to a list of UIDs
//ESpell, Suggests ?improved queries? and spelling fixes.
static void Main(string[] args)
{
//Current Plan
//Get Article IDs from PMC
//Get Journal name, File Name, and Date
//Sort by Journal Name
//Extract tars
//Keep only the Journals & Articles we care about.
//
//Processing ?? Co-Word Occurance ??
//Stop Words = 1:5000 checking every 500,000 words.
string myQuery = "((\"hiv\"[MeSH Terms] OR \"hiv\"[All Fields]) OR (\"acquired immunodeficiency syndrome\"[MeSH Terms] OR (\"acquired\"[All Fields] AND \"immunodeficiency\"[All Fields] AND \"syndrome\"[All Fields]) OR \"acquired immunodeficiency syndrome\"[All Fields] OR \"aids\"[All Fields])) AND hasabstract[text]";
string linkageFileName = "Links.txt";
string articleFileName = "Articles.csv";
//myQuery = "( \"hiv\"[MeSH Terms] OR \"hiv\"[All Fields]) OR (\"acquired immunodeficiency syndrome\"[MeSH Terms] OR (\"acquired\"[All Fields] AND \"immunodeficiency\"[All Fields] AND \"syndrome\"[All Fields]) OR \"acquired immunodeficiency syndrome\"[All Fields] OR \"aids\"[All Fields])";
eUtils.eUtilsServiceSoapClient utilServ = new eUtils.eUtilsServiceSoapClient();
string database = null;
//database = "PMC";
eUtils.eSearchResult searchResult = GetSearchResults(database, myQuery, utilServ);
TextWriter LinksOutputFile = new StreamWriter(linkageFileName);
TextWriter ArticleOutputFile = new StreamWriter(articleFileName);
ArticleOutputFile.WriteLine("PMID,Date,Title,Abstract,MeSH,Citation,Grant");
int Total = int.Parse(searchResult.Count);
while (searchResult != null)
{
if (int.Parse(searchResult.RetStart) > 1000)
break;
foreach (string paper in searchResult.IdList)
{
#region Article Links
eUtils.eLinkResult linkResults = GetLinkResults(paper, utilServ);
//if (linkResults.LinkSet.Length != 0)
//{
LinksOutputFile.Write(paper);
foreach (var a in linkResults.LinkSet)
foreach (var b in a.LinkSetDb)
foreach (var c in b.Link)
if (c.Id.Value != paper && b.DbTo.ToLower() == "pubmed")
LinksOutputFile.Write(" " + c.Id.Value);
LinksOutputFile.WriteLine();
//}
//else
//{
// //No References!?!?
// //Should probably remove from Dataset.
// throw new ArgumentNullException("linkResults", "PubMED ID: " + paper + " has no references, which is improbable to impossible. Throwing Error");
//}
#endregion
#region Article Title, Abstract, and Date
eFetch_pubmed.eUtilsServiceSoapClient fetchServ = new eFetch_pubmed.eUtilsServiceSoapClient();
eFetch_pubmed.eFetchRequest fetchRequest = new eFetch_pubmed.eFetchRequest();
fetchRequest.id = paper;
//fetchRequest.id = string.Join(",", Results);
eFetch_pubmed.eFetchResult fetchResult = fetchServ.run_eFetch(fetchRequest);
for (int i = 0; i < fetchResult.PubmedArticleSet.Length; i++)
{
eFetch_pubmed.PubmedArticleType art = null;
//eFetch_pubmed.PubmedBookArticleType book = null;
if (fetchResult.PubmedArticleSet[i] is eFetch_pubmed.PubmedArticleType)
{
art = (eFetch_pubmed.PubmedArticleType)fetchResult.PubmedArticleSet[i];
if (art.PubmedData.History == null)
{
art.PubmedData.History = new eFetch_pubmed.PubMedPubDateType[1];
art.PubmedData.History[0] = new eFetch_pubmed.PubMedPubDateType();
art.PubmedData.History[0].Year = "N/A";
art.PubmedData.History[0].Month = string.Empty;
art.PubmedData.History[0].Day = string.Empty;
}
if(art.MedlineCitation.Article.ArticleTitle != null)
ArticleOutputFile.Write(string.Join(",",
art.MedlineCitation.PMID.Value,
art.PubmedData.History[0].Year + " " + art.PubmedData.History[0].Month + " " + art.PubmedData.History[0].Day,
//art.MedlineCitation.Article.ArticleDate != null ? art.MedlineCitation.Article.ArticleDate[0].Year + " " + art.MedlineCitation.Article.ArticleDate[0].Month + " " + art.MedlineCitation.Article.ArticleDate[0].Day : "N/A",
'"' + art.MedlineCitation.Article.ArticleTitle.Value.Replace("\"", "''") + '"',
'"' + art.MedlineCitation.Article.Abstract.AbstractText[0].Value.Replace("\"", "''") + '"',
art.MedlineCitation.MeshHeadingList != null ? '"' + string.Join(" ", art.MedlineCitation.MeshHeadingList.Select(o => o.DescriptorName.Value)) + '"' : "N/A",
art.MedlineCitation.CitationSubset != null ? string.Join(" ", art.MedlineCitation.CitationSubset) : "N/A",
(art.MedlineCitation.Article.GrantList != null && art.MedlineCitation.Article.GrantList.Grant != null) ?
string.Join(" ", art.MedlineCitation.Article.GrantList.Grant.Select(o => o.Agency)) : "N/A"));
else if(art.MedlineCitation.Article.Journal.Title != null)
ArticleOutputFile.Write(string.Join(",",
art.MedlineCitation.PMID.Value,
art.PubmedData.History[0].Year + " " + art.PubmedData.History[0].Month + " " + art.PubmedData.History[0].Day,
'"' + art.MedlineCitation.Article.Journal.Title.Replace("\"", "''") + '"',
'"' + art.MedlineCitation.Article.Abstract.AbstractText[0].Value.Replace("\"", "''") + '"',
art.MedlineCitation.MeshHeadingList != null ? '"' + string.Join(" ", art.MedlineCitation.MeshHeadingList.Select(o => o.DescriptorName.Value)) + '"' : "N/A",
art.MedlineCitation.CitationSubset != null ? string.Join(" ", art.MedlineCitation.CitationSubset) : "N/A",
(art.MedlineCitation.Article.GrantList != null && art.MedlineCitation.Article.GrantList.Grant != null) ?
string.Join(" ", art.MedlineCitation.Article.GrantList.Grant.Select(o => o.Agency)) : "N/A"));
ArticleOutputFile.WriteLine();
//foreach (var v in art.MedlineCitation.Article.Abstract.AbstractText)
//Console.WriteLine(v.Value);
//foreach(string a in art.MedlineCitation.CitationSubset)
//Console.WriteLine(a);
}
else if (fetchResult.PubmedArticleSet[i] is eFetch_pubmed.PubmedBookArticleType) { }
else if (fetchResult.PubmedArticleSet[i] is eFetch_pubmed.PubmedBookDataType) { }
else if (fetchResult.PubmedArticleSet[i] is eFetch_pubmed.PubmedDataType) { }
else if (fetchResult.PubmedArticleSet[i] is eFetch_pubmed.PubMedPubDateType) { }
else if (fetchResult.PubmedArticleSet[i] is eFetch_pubmed.PubMedPubDateTypePubStatus) { }
}
#endregion
}
searchResult = GetSearchResults(database, myQuery, utilServ, searchResult);
}
LinksOutputFile.Flush();
LinksOutputFile.Close();
LinksOutputFile.Dispose();
ArticleOutputFile.Flush();
ArticleOutputFile.Close();
ArticleOutputFile.Dispose();
}
private void btnGo2_Click(object sender, EventArgs e)
{
// example retrieves links from Nuccore for to Protein GI 48819,7140345
try
{
eUtils.eUtilsServiceSoapClient serv = new eUtils.eUtilsServiceSoapClient();
string[] id = { "48819,7140345" };
// call NCBI ELink utility
eUtils.eLinkRequest req = new eUtils.eLinkRequest();
req.db = "protein";
req.id = id;
req.dbfrom = "nuccore";
eUtils.eLinkResult res = serv.run_eLink(req);
// results output
textBox1.Text = "";
for (int i = 0; i < res.LinkSet.Length; i++)
{
textBox1.Text += "Links from " + res.LinkSet[i].DbFrom +
" to " + res.LinkSet[i].LinkSetDb[0].DbTo + "\r\n";
textBox1.Text += " " + res.LinkSet[i].DbFrom + " id(s): ";
for (int k = 0; k < res.LinkSet[i].IdList.Length; k++)
{
textBox1.Text += res.LinkSet[i].IdList[k].Value + " ";
}
textBox1.Text += "\r\n";
textBox1.Text += " " + res.LinkSet[i].LinkSetDb[0].DbTo + " id(s): ";
for (int k = 0; k < res.LinkSet[i].LinkSetDb[0].Link.Length; k++)
{
textBox1.Text += res.LinkSet[i].LinkSetDb[0].Link[k].Id.Value + " ";
}
textBox1.Text += "\r\n----------------------\r\n";
}
}
catch (Exception eee)
{
textBox1.Text = eee.ToString();
}
}
private void btnGo_Click(object sender, EventArgs e)
{
// O utilitário eInfo retorna uma lista dos bancos de dados disponíveis
try
{
eUtils.eUtilsServiceSoapClient serv = new eUtils.eUtilsServiceSoapClient();
// chama o NCBI EInfo utility
eUtils.eInfoResult res = serv.run_eInfo(new eUtils.eInfoRequest());
// results output
this.txtResult.Text = "";
for (int i = 0; i < res.DbList.Items.Length; i++)
{
txtResult.Text += res.DbList.Items[i] + "\r\n";
}
}
catch (Exception ex)
{
txtResult.Text = ex.ToString();
}
}
public string bustaTextual(String termo)
{
StringBuilder builder = new StringBuilder();
StringBuilder builderFinal = new StringBuilder();
string retorno = "";
try
{
eUtils.eUtilsServiceSoapClient serv = new eUtils.eUtilsServiceSoapClient();
// call NCBI ESearch utility
// NOTE: search term should be URL encoded
String inicio = "0";
Boolean teste = true;
int qnt = 0;
int qntfound = 0;
do{
eUtils.eSearchRequest req = new eUtils.eSearchRequest();
req.db = "nuccore";
req.term = termo.Replace(" " ,"+");
req.RetStart = inicio;
req.RetMax = "100000";
eUtils.eSearchResult res = serv.run_eSearch(req);
qntfound = Convert.ToInt32(res.Count);
qnt = qnt + res.IdList.Length;
if (qnt == qntfound)
{
builder.Clear();
int cont = 0;
for (int x = 0; x < res.IdList.Length; x++)
{
if (cont == 500)
{
builderFinal.Append(getStringPost(createURL(builder.ToString(), "nuccore")));
richTextBox1.Text += builderFinal.ToString();
builderFinal.Clear();
builder.Clear();
cont = 0;
}
else
{
builder.Append(res.IdList[x] + ",");
cont++;
}
}
teste = false;
}
else
{
builder.Clear();
int cont = 0;
for (int x = 0; x < res.IdList.Length; x++)
{
if (cont == 500)
{
builderFinal.Append(getStringPost(createURL(builder.ToString(), "nuccore")));
richTextBox1.Text += builderFinal.ToString();
builderFinal.Clear();
builder.Clear();
cont = 0;
}
else
{
builder.Append(res.IdList[x] + ",");
cont++;
}
}
}
inicio = Convert.ToString(Convert.ToInt32(inicio) + 100000);
}while(teste);
MessageBox.Show(Convert.ToString(qntfound) + " Id`s Encontrados e armazanados em Buffer... \n Clique em OK para MOstrar as sequências");
retorno = builderFinal.ToString();
return retorno;
}
catch (Exception eee)
{
textBox1.Text = eee.ToString();
return retorno;
}
}
