With only two characters, you can tailor your game to their roles. I've always enjoyed the opportunity to do this with one- and two-player campaigns.
Consider a class like the druid. In a group with a wizard, thief, undead-smiting cleric, and a paladin, just how much time is that druid going to be spending in the forest doing druid-y forest-protect-y things? Perhaps some, but most is going to be spent underground, fighting undead and such. How great would it be to have the chance to actually make the game entirely focus on the things that are relevant to a character's class?
With a two-player party, you can take the average of their characters' themes and make a campaign that focuses on that:
Have a thief and a bard? Great, play a city game where sneaking, fast-talking, and investigating are the order of the day. You don't have to worry about them being weak in a fight—make the fights few and weak! After all, when all you have is a thief and a bard, two heavily-armed soldiers are scary—you don't have to make the encounters the same as if there were a full role-set in the party.
Have a ranger and a druid? No need for a thief to disarm traps, because really, how many traps are they going to find in the forest? Deadfalls or tiger-pit traps on rare occasions perhaps, but those need to be spotted and avoided, not disarmed. Do they run into a bit of arcane magic that would be trivial to understand/bypass with a wizard in tow? Great! Now it's a significant, mysterious, frightening bit of unearthly enchantment that can be a bigger deal to the story!
Two clerics? Have fun stomping undead, dealing with opposed evil cults, and other religion-y themes!
A wizard plus any-other-class? Excellent, you've got a magical adventure in the future. Wizards are way more interesting when they're not just moving artillery platforms. Throw in enchantments, ancient runes, mysterious libraries, and the machinations of rival wizards!
Tailoring a campaign to a small party is a refreshing, pleasant experience. Make the most of it by really digging into the themes that make the players' characters interesting to them, and you'll have a memorable game without worrying about "balancing" out the encounters to a non-existent "normal" party. This is often quite difficult when running prepared adventures, but when you're homebrewing a campaign as you are, it's easy and a treat to GM.
"A practical man can always make what he wants to do look like a noble sacrifice of personal inclinations to the welfare of the community. I've decided that I've got to be practical myself, and that's one of the rules. How about breakfast?" The Pirates of Ersatz, Murray Leinster
From your question I noticed a few things. Nominally, I completely agree with @mxyzplk's answer, so this should be in the way of an addendum.
It sucks to be the leader
In a RPG, it just completely sucks to be the leader. Most players when confronted with a plan, remember about fifteen percent of it for the first fifteen minutes. But they'll certainly remember when you deviate. Leaders get no additional responsibility and no perquisites, but they get all the blame.
In the military this is mitigated with the clear distinction between commissioned and non-commissioned officers. Not least because the isolation provides both support structures and necessary emotional distance (to a degree, of course). Being "elected" leader, especially with the pack dynamics of typical werewolf games is an extremely dubious honour that I'd flatly reject.
The fact that while you may be leader in character but not dominant over the player group makes things even stickier. You need to assert authority within the realm of the narrative without actually having that authority in reality. Again, something that will cause friction and resentment any way you cut it.
Depressing environments bleed emotions into play
The world of darkness does what it says on the tin. Having played in a horror game myself recently, the iconic themes of the world of darkness do not make for "happy" or, for that matter, validating game experiences in the main. (And, if they do, it's a violation of genre.) When you are faced with the stresses of being "leader" which are compounded by the stressors of the philosophies baked into the setting, no wonder you're having a rough time.
Some solutions:
On leadership:
Fundamentally, a gaming group is a relationship. Bad relationships that do not provide validation are a drain on mental and emotional resources. When they don't work, cut them off or change them. In your case, I'd play a game that's a bit lighter in tone and focus: a nice traditional dungeon crawl or similar heroic fantasy.
I'd also reject the leader role for all the reasons I outlined above. Or, if they force it upon you, demand the perquisites and authority that is concomitant with it: they can't have it both ways.
On the group:
I've found that group character creation creates a far more cohesive group. By having entangled backstories, the group can draw upon a deeper understanding of each others' characters, creating the basis for empathy and respect within the characters, instead of the necessary simulacrum imposed by players.
By articulating desired tropes, a "palette" (as Microscope) calls it, before the game begins, you'll be able to shape the narrative of the group in directions that you want to play. This allows you to avoid the nominally depressive tropes that come default with the setting (not limited to world of darkness) and describe a source for future characters to connect with the current group. Replacement characters, if they tie into the shared narrative, will continue to maintain the tropes and social trust.
Be practical:
As players, we shape our narratives to an amazing degree. Emulate Bron Hoddan in the Pirates of Ersatz. While playing, you will be aware of the desired practical outcome that will provide validation and satisfy your personal goals. With that outcome in mind, you then frame it in terms that suit both your character's narrative and the expected narratives of the other players such that they will act to reinforce your framing and thereby your outcome. If you fight their narrative control by "being a loner," it is difficult to achieve your own goals. If you help them work as a team and appear to sacrifice nobly on their behalf while executing your own goals... the entire process is smoother and more effective.
Note that I am not saying to lie. Instead, consider the causal constructions of your actions, the explanations for those actions to be an aspect of the role * separate* from the actions themselves. By manipulating the framing as well as the actions, you can provide the necessary hooks for the other players to support your version of reality, rather than rejecting it and, by extension, you.
Postscript
Looking at your comments to other questions, you should absolutely give this group two last tries. In the first trial (of one or two games), try a heroic romp where you can be "Big Damn Heroes." Require the players who need the spotlight be leader. In the second trial (again of one or two games), try a game where players can intrigue against each other (I'd recommend Ars Magica, but then again I recommend it for most things. Most games support PvP intrigue quite ably.) If neither game provides the validation you need and the spotlight the other players need, move on. Before you do anything, take a month break, sit down, relax, and try to game with some strangers. I'm pretty sure that if you go looking for games in the chat section of this site... someone will oblige. For more on the framing problem, I'd quite recommend Rule 34 by Stross, as it describes it in a delicious narrative context.
Best Answer
Roles Really Aren’t That Important in 3.5
To begin, spells are the most powerful class feature in the game. Dungeons & Dragons 3.5 doesn’t really care much about roles: you will be more powerful the more magic you have. You will never be more powerful going for a non-magic class, even if the rest of your group is already magical. So I call this party a good idea on that account. Unfortunately, you do have some pretty severe power disparity here...
Tier-1s
Wizard and Druid are two of the most powerful classes in the game. All else equal, they can seriously handle a lot more than a group of, say, Monk, Paladin, Fighter, Rogue, and Healer could, at all but the lowest levels of optimization.
Tier-3-ishes
Duskblade and Dread Necromancer are good classes, certainly capable of holding their own. Both are not-awful without spells (Duskblade being nearly as good as a Fighter would have been, and Dread Necromancer having an army of minions), too.
Tier-4-ish
Spellthief is a pretty weak class, but it’s pretty cool. There’s a Trickster variant from Dragon vol. 353 that’s pretty good (lose all but the first 1d6 of Sneak Attack, a fair few skills, and Trapfinding for Bard-like spell progression and adding the Bard spells to your list), though it may make the character less “theif-y” than the player wants.
If possible, you should make Godsblood Spelltheft available to the Spellthief – select Deities or Domains that are appropriate for your setting. This feat gives the Spellthief some nice reliability: if the spell he steals isn’t so great, he can always use one of his Domain spells. It helps quite a bit.
The Spellthief is also the only character with Use Magic Device on his class skill list, and only the Dread Necromancer also uses Charisma. This makes him pretty ideal for the party’s wand-wielder. In particular, your party lacks access to Cleric spells, which are quite good. Wands require a static DC 20 UMD check, which is possible to hit pretty consistently by mid-low levels. Not all spells are good choices for wands, but judicious use of them can give the Spellthief a lot of options.
You actually do have the four roles...
Despite the fact that you don’t really need them, you do in fact have the four traditional roles. The Duskblade is a warrior-type, the Wizard is your arcanist, Druids can heal quite well with lesser vigor (I recommend he gets a wand of it for keeping everyone topped up between fights), and the Spellthief has Trapfinding and the like. Dread Necromancers provide a pretty good “fifth class” here – minions can be whatever you need them to be, and he’s got decent curses and battlefield control.
Of course... the Druid or the Wizard could very easily handle front-lining (Wild Shape, polymorph, etc.), or trap-killing (summons are great for that), or minions (Animal Companion, Handle Animal, summons), or... just about anything else...
Anti-magic Field
Since this got brought up in a comment, some deals with anti-magic field.
It’s a high-level spell.
It’s always centered on the caster.
It’s got a really small radius.
A caster casts anti-magic field on himself, and he’s done.1 Even if a warrior gets an Anti-magic Torc, activating it kills all his magic items, and now he’s in a lot of trouble too. And in both cases, they’ve got to get within range of a spellcaster, activate the AMF (or somehow get in range without magic, but a spellcaster who lets that happen doesn’t deserve the name), and then do something before the spellcaster simply leaves. That radius is way too small to easily pin someone in it, and actions are working against you here.
Casters should be aware of AMF, and build for it. Conjuration (Creation) spells that have SR: No fly right through an AMF – the Wizard’s got a lot of those in Spell Compendium in the form of the orb of spells. The Druid’s Animal Companion, and all of the Dread Necromancer’s minions, can operate more-or-less unaffected by an AMF. The Duskblade does have full BAB and good HP and armor, which means he’s about as good as any other warrior in an AMF. The Spellthief is really the only one who will be largely sidelined by an AMF.
Dead Magic Zone
This is like an AMF, without any of the problems of an AMF. Worse, it’s not even associated with an enemy – it just is. Just, don’t use these. They’re pretty bad for a normal game (arbitrary and asymmetric crippling of characters, with absolutely no recourse), and for this party it would just be a jerk thing to do. At that point you might as well just “rocks fall, everyone dies.”
1 For the purposes of this discussion, I am ignoring the Cheater of Mystra. If you’re seriously playing at that level of optimization, you’re already well out of my league and I cannot help you.