++moritz for the idea of succinctly expressing this in SQL. (I will address it in Perl in a later post).

Perhaps I read the OP differently, but I think that this SQL, as given, will not produce the desired results.

Here is my test setup, using SQLite (tested on Win32 and Linux):

sqlite3 PM_617794.db
DROP TABLE IF EXISTS pdbs;
CREATE TABLE pdbs(
  PdbId       TEXT,
  MetalAtm    TEXT,
  MetalName   TEXT,
  MetalNumber INTEGER,
  MetalChn    TEXT,
  AtmName     TEXT,
  ResName     TEXT,
  ResNumber   INTEGER,
  ResChn      TEXT,
  Distance    NUMERIC
);
INSERT INTO pdbs VALUES('1nc7','MG','MG',1501,'-','CL','CL',1401,'-',3
+.065);
INSERT INTO pdbs VALUES('1nc7','MG','MG',1501,'-','CL','CL',1402,'-',2
+.660);
INSERT INTO pdbs VALUES('1pex','CA','CA', 503,'-','CL','CL', 501,'-',2
+.895);
INSERT INTO pdbs VALUES('1qhu','NA','NA',   3,'-','CL','CL',   1,'-',3
+.096);
INSERT INTO pdbs VALUES('1qjs','NA','NA', 513,'A','CL','CL', 511,'A',3
+.096);
INSERT INTO pdbs VALUES('1rtg','CA','CA',   2,'-','CL','CL',   1,'-',3
+.079);
INSERT INTO pdbs VALUES('1xj6','NA','NA',5002,'-','CL','CL',5001,'-',2
+.628);
INSERT INTO pdbs VALUES('2caj','NI','NI',1151,'B','CL','CL',1150,'B',3
+.551);
INSERT INTO pdbs VALUES('2oa9','CD','CD', 911,'-','CL','CL', 800,'-',2
+.291);
INSERT INTO pdbs VALUES('2oa9','CD','CD', 913,'-','CL','CL', 801,'-',2
+.403);
.mode column pdbs
.width 4 2 2 4 1 2 2 4 1 5
[download]

Here is your query, with the FROM clause added, and the $columns resolved to '*':

SELECT   *, COUNT(PdbId) as c
FROM     pdbs
GROUP BY PdbId
HAVING   c >= 2
;
[download]

Output:

1nc7  MG  MG  1501  -  CL  CL  1402  -  2.66   2
2oa9  CD  CD  913   -  CL  CL  801   -  2.403  2
[download]

There are two problems here:

1. Any GROUP BY clause can produce only *one* record per unique grouped-by field. The OP needs *all* the records which have groupings of more than one record. To achieve this, you must either use
   1. a sub-Select, or
   2. two Selects with a temp table storing the results of the first Select for use by the second Select.
   I don't see how it can be done with a single simple Select.
2. Your SQL would produce a record if a PdbId had two records of *any* values of ResName. The sample data does not happen to have this scenario, but the code should not depend on that. The issue can be resolved by
   1. filtering with a WHERE clause; WHERE ResName = 'CL', or
   2. (more maintainably and flexibly) by adding ResName to the GROUP BY clause, and also filtering with WHERE.
   The OP *could* be read as using CL only as a representative of matching values; if so, then the WHERE is not needed, but the addition to GROUP BY becomes mandatory.

Here is your SQL, modified as 1a and 2b above:

SELECT     pdbs.*
FROM       pdbs JOIN (
  SELECT   PdbId, ResName, COUNT(*) as c
  FROM     pdbs
  GROUP BY PdbId, ResName
  HAVING   c >= 2
) AS       counter
ON         pdbs.PdbId   = counter.PdbId
AND        pdbs.ResName = counter.ResName
WHERE      pdbs.ResName = 'CL'
;
[download]

Outputs:

1nc7  MG  MG  1501  -  CL  CL  1401  -  3.065
1nc7  MG  MG  1501  -  CL  CL  1402  -  2.66
2oa9  CD  CD  911   -  CL  CL  800   -  2.291
2oa9  CD  CD  913   -  CL  CL  801   -  2.403
[download]