Fixed a bunch of bugs. Filters now are called once for all categories…

…. Added more program options for collate sub-command.

Actions/SearchAction.cs

@@ -31,11 +31,11 @@ public int Setup(SearchOptions options)
 
             if (patternName == null)
             {
-                Console.Error.WriteLine($"ERROR: Cannot find pattern name in {options.Pattern.FullName}");
+                Logger.Error($"Cannot find pattern name in {options.Pattern.FullName}");
                 return 1;
             }
 
-            Console.WriteLine($"Found pattern name {patternName}");
+            Logger.Info($"Found pattern name {patternName}");
 
             // Try to find the pattern CSV files in the pattern directory
             var patternCsvs = new Dictionary<Category, FileInfo>();
@@ -54,12 +54,12 @@ public int Setup(SearchOptions options)
 
             if (patternCsvs.Count == 0)
             {
-                Console.Error.WriteLine($"ERROR: No pattern found in {options.Pattern.FullName}");
+                Logger.Error($"No pattern found in {options.Pattern.FullName}");
                 return 2;
             }
 
             if (errorList.Count > 0)
-                Console.Error.WriteLine($"WARN: Some patterns are missing: {string.Join(", ", errorList)}");
+                Logger.Warning($"Some patterns are missing: {string.Join(", ", errorList)}");
 
             // Finally, set the model for running this action
             Parameters = new SearchParameters(
@@ -81,12 +81,13 @@ public int Run()
             // Must have been set in the Setup method
             Debug.Assert(Parameters != null);
 
-            Console.WriteLine($"Using {Parameters.Similarity.Type} similarity measure");
+            // Print out key parameters
+            Logger.Info($"Using {Parameters.Similarity.Type} similarity measure");
 
             Parallel.ForEach(Parameters.PatternCsvs, o =>
             {
                 var (category, patternFile) = o;
-                Console.WriteLine($"Searching in category {category}");
+                Logger.Info($"Searching in category {category}");
 
                 // Load the pattern vector for matching
                 var patternResDict = File.ReadAllLines(patternFile.FullName)
@@ -142,7 +143,7 @@ public int Run()
                     count++;
                 }
 
-                Console.WriteLine($"Generating top {Parameters.ResultCount} matches out of {count} {category} vectors");
+                Logger.Info($"Generating top {Parameters.ResultCount} matches out of {count} {category} vectors");
 
                 // Prepare the category specific directory for storing output
                 string categoryDir = Path.Combine(Parameters.OutputDir.FullName, category.ToString());
@@ -202,7 +203,7 @@ private static void CopyBestConformation(string inputPdbqt, string outputPdbqt,
             int startIndex = text.IndexOf($"MODEL {confId,8}");
             if (startIndex == -1)
             {
-                Console.Error.WriteLine($"ERROR: Failed to find conformation {confId} in {inputPdbqt}");
+                Logger.Error($"Failed to find conformation {confId} in {inputPdbqt}");
                 return;
             }
 

Enums/RowOrdering.cs

@@ -0,0 +1,8 @@
+namespace mccsx
+{
+    internal enum RowOrdering
+    {
+        score,
+        sequence,
+    }
+}

Exceptions/FilterException.cs

@@ -0,0 +1,20 @@
+using System;
+using System.Runtime.Serialization;
+
+namespace mccsx
+{
+    [Serializable]
+    public class FilterException : Exception
+    {
+        public FilterException() { }
+        public FilterException(string message) : base(message) { }
+        public FilterException(string message, Exception inner) : base(message, inner) { }
+        public FilterException(string message, string filterName) : base(message) => FilterName = filterName;
+        public FilterException(string message, string filterName, Exception inner) : base(message, inner) => FilterName = filterName;
+        protected FilterException(
+          SerializationInfo info,
+          StreamingContext context) : base(info, context) { }
+
+        public string? FilterName { get; }
+    }
+}

Extensions/ExcelFluentExtensions.cs

@@ -9,9 +9,9 @@ namespace mccsx.Extensions
 {
     public static class ExcelFluentExtensions
     {
-        public static SpreadsheetDocument OpenXlsxFile(this string filepath)
+        public static SpreadsheetDocument OpenXlsxFile(this string filepath, bool forceOverwrite)
         {
-            if (!File.Exists(filepath))
+            if (forceOverwrite || !File.Exists(filepath))
             {
                 // Create a spreadsheet document by supplying the filepath.
                 // By default, AutoSave = true, Editable = true, and Type = xlsx.

Extensions/MeasureExtensions.cs

@@ -17,7 +17,7 @@ public static IVectorDistanceMeasure SimilarityMeasure(this Measure measure)
 
         public static IVectorDistanceMeasure DistanceMeasure(this Measure measure)
         {
-            var simClassType = EnumAnnotationHelper<Measure>.GetAttribute<MeasureImplAttribute>(measure).SimilarityClass;
+            var simClassType = EnumAnnotationHelper<Measure>.GetAttribute<MeasureImplAttribute>(measure).DistanceClass;
             var obj = Activator.CreateInstance(simClassType);
             Debug.Assert(obj != null && obj is IVectorDistanceMeasure);
             return (obj as IVectorDistanceMeasure)!;

Filters/IndexFilter.cs

@@ -14,7 +14,8 @@ internal class IndexFilter
 
         public IndexFilter(string[] lines)
         {
-            Debug.Assert(lines.Length > 0);
+            if (lines.Length < 2)
+                throw new FilterException($"Insufficient lines", "index");
 
             string[][] fieldLines = lines
                 .Select(o => o.Split(" \t".ToCharArray(), StringSplitOptions.RemoveEmptyEntries))
@@ -30,7 +31,7 @@ public IndexFilter(string[] lines)
             }
             else
             {
-                throw new FilterColumnException("no enough columns");
+                throw new FilterException("Insufficient columns");
             }
         }
 

Filters/StateFilter.cs

@@ -13,7 +13,8 @@ internal class StateFilter
 
         public StateFilter(string[] lines)
         {
-            Debug.Assert(lines.Length > 0);
+            if (lines.Length < 2)
+                throw new FilterException($"Insufficient lines", "state");
 
             string[]? headers = lines[0].Split(" \t".ToCharArray(), StringSplitOptions.RemoveEmptyEntries);
 
@@ -38,6 +39,8 @@ public StateFilter(string[] lines)
                     if (nameEnd == 0)
                     {
                         int idx = line.IndexOf(' ');
+                        if (idx == -1)
+                            idx = line.IndexOf('\t');
                         if (idx != -1)
                         {
                             nameBegin = 0;
@@ -47,7 +50,7 @@ public StateFilter(string[] lines)
 
                     if (nameEnd == 0)
                     {
-                        throw new FilterColumnException($"no enough columns on line: {line}");
+                        throw new FilterException($"Insufficient columns on line: {line}");
                     }
 
                     _data[line[nameBegin..nameEnd]] = line[nameEnd..].Trim();
@@ -56,7 +59,7 @@ public StateFilter(string[] lines)
             }
             else
             {
-                throw new FilterColumnException("no enough columns");
+                throw new FilterException("Insufficient columns");
             }
         }
 

Helpers/Logger.cs

@@ -0,0 +1,26 @@
+using System;
+
+namespace mccsx.Helpers
+{
+    internal static class Logger
+    {
+        public static void Info(string message)
+        {
+            Console.WriteLine(message);
+        }
+
+        public static void Warning(string message)
+        {
+            Console.ForegroundColor = ConsoleColor.Yellow;
+            Console.Error.WriteLine($"WARN: {message}");
+            Console.ResetColor();
+        }
+
+        public static void Error(string message)
+        {
+            Console.ForegroundColor = ConsoleColor.Red;
+            Console.Error.WriteLine($"ERROR: {message}");
+            Console.ResetColor();
+        }
+    }
+}

Models/InputVectors.cs

@@ -93,13 +93,13 @@ public static InputVectors FromRawRecvData(
             // Verify consistency of index names
             foreach (var recv in vecData)
             {
-                if (recv.IndexName == null)
+                if (indexName == null)
                 {
                     indexName = recv.IndexName;
                 }
                 else if (indexName != recv.IndexName)
                 {
-                    errorMessages.Add($"index name '{recv.IndexName}' different from previous ones '{indexName}' for {recv.VectorName}");
+                    errorMessages.Add($"Index name '{recv.IndexName}' different from previous ones '{indexName}' for {recv.VectorName}");
                 }
             }
 
@@ -114,7 +114,7 @@ public static InputVectors FromRawRecvData(
             var summaryFields = vecData.ToDictionary(o => o.VectorName, o => o.SummaryFields);
 
             // Create the instance
-            return new InputVectors
+            var obj = new InputVectors
             (
                 vecData.Select // Column vectors
                 (
@@ -137,6 +137,10 @@ public static InputVectors FromRawRecvData(
                 summaryFields, // Summary fields
                 errorMessages // Error messages
             );
+
+            Debug.Assert(obj.ColumnCount == vecData.Count);
+
+            return obj;
         }
 
         private static void SummarizeResidues(
@@ -200,7 +204,7 @@ private static void SummarizeResidues(
             residueInfoByIndex = chainDictByIndex.Keys.ToDictionary(idx => idx, idx => new ResidueInfo
             (
                 string.Join('/', chainDictByIndex[idx].OrderBy(p => p)),
-                string.Join('/', resNameDictByIndex[idx].Select(p => p.GetShortName()).OrderBy(p => p)),
+                string.Join('/', resNameDictByIndex[idx].Select(p => p.GetShortName().ToUpper()).OrderBy(p => p)),
                 string.Join('/', resSeqDictByIndex[idx].OrderBy(p => p)),
                 resSeqDictByIndex[idx].Count == 1 ? resSeqDictByIndex[idx].First() : (int?)null,
                 idx,
@@ -211,7 +215,7 @@ private static void SummarizeResidues(
             residueInfoBySeq = chainDictBySeq.Keys.ToDictionary(seq => seq, seq => new ResidueInfo
             (
                 string.Join('/', chainDictBySeq[seq].OrderBy(p => p)),
-                string.Join('/', resNameDictBySeq[seq].Select(p => p.GetShortName()).OrderBy(p => p)),
+                string.Join('/', resNameDictBySeq[seq].Select(p => p.GetShortName().ToUpper()).OrderBy(p => p)),
                 seq.ToString(),
                 seq,
                 $"{string.Join('/', resNameDictBySeq[seq].Select(p => p.GetCode()).OrderBy(p => p))}{seq}",
@@ -266,16 +270,23 @@ public Rect GetFormattedReport(
                 }
 
                 // Header row 1
-                var headerRow1 = RawRecvData.CsvColumnHeaders
-                    .Append(StateName)
-                    .Concat(columnKeys)
+                IEnumerable<string?> headerRow1 = RawRecvData.CsvColumnHeaders;
+
+                if (IndexName != null)
+                    headerRow1 = headerRow1.Append(IndexName);
+
+                headerRow1 = headerRow1.Concat(columnKeys)
                     .Concat(stateCounts.Select(o => $"Average({o.Count})"))
                     .Append($"Average({ColumnCount})");
 
                 // Header row 2
-                var headerRow2 = RawRecvData.CsvColumnHeaders
-                    .Select(o => (string?)null)
-                    .Append($"{StateName}->")
+                IEnumerable<string?> headerRow2 = RawRecvData.CsvColumnHeaders.SkipLast(1)
+                    .Select(o => (string?)null);
+
+                if (IndexName != null)
+                    headerRow2 = headerRow2.Append(null);
+
+                headerRow2 = headerRow2.Append($"{StateName}->")
                     .Concat(columnKeys.Select(o => GetColumn(o).Tag))
                     .Concat(stateCounts.Select(o => o.State))
                     .Append("All");
@@ -306,9 +317,9 @@ public Rect GetFormattedReport(
             int width = left + ColumnCount + (StateName != null ? stateCounts.Length : 0);
             int height = top + RowCount + RawRecvData.SummaryRowHeaders.Count;
 
-            // Build the data rows, in the ascending lexical ordering
+            // Build the data rows, in the ascending ordinal ordering
             dataRows = Rows
-                .OrderBy(vec => vec.Name)
+                .OrderBy(vec => ResidueInfo[vec.Name].ResidueSeq)
                 .Select((vec, i) =>
                 {
                     var resInfo = ResidueInfo[vec.Name];
@@ -371,7 +382,7 @@ public Rect GetFormattedReport(
         /// <returns></returns>
         private static string GetRowAverage(int rowNum, int colNum, int cols)
         {
-            return $"=AVERAGE({ExcelColumnHelper.ToCellRef(colNum, rowNum)}:{ExcelColumnHelper.ToCellRef(cols, rowNum + cols - 1)})";
+            return $"=AVERAGE({ExcelColumnHelper.ToCellRef(colNum, rowNum)}:{ExcelColumnHelper.ToCellRef(colNum + cols - 1, rowNum)})";
         }
 
         public IReadOnlyDictionary<string, string>? GetRowTags()